Williams Stargate cocktail - project repair

[tmcw]

Thanks for the replies...

I'll order the cap and some socket strips while I'm at it, I think I'm going to need them!

Mod is done correctly, meant I was getting the +12v reg at the header, and +5v at the header and pins on the RAMs, so my power board and connections are definitely better than they were.

I'm in Ireland, so I presume the power transformer is correctly jumpered for our voltage; I think it came from the UK because there were UK coins in the coin box. The internal fan was disconnected because it's a 110v fan.

I did a little bit of testing with the RAM chips today, mixing up the original 8264A's and newer 4164's, and starting with a 1-3-1 error, and moving chips around or replacing, I was able to walk the error to 1-3-2, down a few spaces to 1-3-8, then over to 1-1-1 and then 1-1-2, but couldn't seem to get past 1-1-2.

However, I'm wondering about the behavior of the RAM checking sequence, does it start at 1-3-1, to 1-3-2, etc, to 1-3-8, then 1-2-1 to 1-2-8, then 1-1-1 to 1-1-8? Also, assuming everything is good electronically, except RAM chips, will it only display the first RAM chip that it finds as faulty? The reason I ask is because sometimes I might see a 1-3-1 error, then if I press a chip, or even flex the board a little, I might see (most commonly) a 1-3-6 or 1-3-7 error. If it should only display the first chip with error that it encounters, then it looks like I might have flakey sockets here. Is that a fair assumption, or anything else I need to consider? Ribbon connection to ROM board? I don't notice any changes in the RAM error number when I move the ribbon connection about.

My other question is about testing my RAM chips. I don't have a dedicated chip tester, but is it possible to use the board to test them? For example, if the first chip tested is 1-3-1 position, and if I have just one chip in the board at that location, and get a 1-3-1 error, then the chip is bad. If I get a 1-3-2 error, then the chip at 1-3-1 is good. Is it as simple as that, or are there other things to consider, besides ensuring that the socket is good and the voltage getting to the chip is high enough?

 

[YellowDog]

Assuming that the sockets are good (and there are a number of things that can turn sockets bad) the best way of testing for RAM failures is to move the indicated RAM chip down in the bank. If it is chip 33 (1-3-3 failure) I move it to position 32 and so on. If it is at 31 I move it to position 21.

The RAM test stops on the first bad chip. The tests go from bank 3 to bank 2 to bank 1 and from chip 1 to chip 8.

So by moving it down the bank, you are moving a chip that passed up (later in the test) and swapping a questionable chip down (earlier in the test).

The other thing to do is to pull all the chips and put them in a bowl. Borrow a couple of your wife's (or girlfriend's) nail polish in different colors. Then put all the chips in and power up. Mark the first bad chip with the first color. Pull it, set it aside and replace it with a different chip. Once you have a complete set that is working, replace the ones you marked, one at a time and see if they will pass. If it passes, move it to a different position and test it again.

Because the rug test does not actually test every single bit, it is possible to have single bit errors that are not detected, unless you move the chip.

ken

 

[tmcw]

Thanks for that, that's a great explanation, exactly what I needed to know.

I tried it out, but I can't seem to get beyond a 1-3-7 error now. I tried moving the chip in 3-7 to 3-6, but the error is staying at 3-7. I tried moving a couple of chips between 3-1 and 3-5 which are supposedly good, to 3-7, but still the error remains at 3-7.

It's strange, I changed a chip at 1-2, as that was where I was getting stuck before, and this resulted in getting a 3-1 error again. When I changed it back, I went back to getting a 3-7 error, even though I didn't touch that chip.

I tested all the sockets on the board, RAM, MPU, decoder ROMs, and the other two (4L and 6C) that look to be post-factory, and they all check out ok.

So where do I go from here?

 

[tmcw]

Ok, had some time to spend on this today, so was following the cpu logic check from the Stargate/Robotron document.

Machine still gives a 1-3-7 error, and a constantly cycling "rug" wipe.

First few steps in the flow diagram checked out ok until I got to check 1D, 2D and 2G (MPU Address Buffers).

1D was strobing ok.

Pins 5, 7 and 9 on 2D, and 3 and 5 on 2G were only strobing when the "rug" was wiping; while not wiping, they were stuck high. All other pins were ok.

Going onto the next step in the flow diagram, 1G (MPU Data Buffers), all pins except 19 were strobing ok. Pin 19 was stuck low.

Anyone familiar enough with these boards to help me interpret what's going on, and what to check or do next?

 

[tmcw]

Also, I currently have the 3 AA batteries and the battery holder off the board, awaiting a replacement coin-cell holder.

This isn't essential to have in-place to be able to trouble-shoot the board, is it?

 

[YellowDog]

No. All that will happen is that you will have to reset the CMOS every time via the advance switch inside the coin door.

ken

 

[tmcw]

... The RAM test stops on the first bad chip. The tests go from bank 3 to bank 2 to bank 1 and from chip 1 to chip 8....

ken

Had some correspondence with Dave from robotron-2084.co.uk, and he said that the RAM test goes bank 3 (1->8), bank 1 (1->8) then bank 2 (1->8).

Anyhow, I did some major testing with the RAM chips, and have the error walked out to 1-1-1, but can't budge it off this position. When I switched chips from bank 3 to bank 1, the error remains at bank 1.

I then changed the socket at that position, but I still get error 1-1-1.

Tested all the traces off this chip, and they all go exactly where they should go. Also, the intermittent error I was having, where if I flexed the board, I could see the error move about (from 1-1, back to 3-6 or 3-7, and sometimes to 3-1), that behaviour has disappeared since I changed the socket at 1-1. Two of the solder points under this socket did look a bit "porous" or grainy, so maybe that was the issue there.

Anyone have any suggestions of where to go next with this?

 

[YellowDog]

If you have a logic probe, you can check to make sure the data lines are pulsing. In particular, you need to check pin 2 on the RAM chips. If it is not pulsing, then you need to check the chip at 1H which on a Stargate is a 74LS374. If your MPU board was swapped out for a later board, this might be a 74LS373. Another potential chip to test is 4E, a 74153 that maps the different address buses to the memory address bus.

ken

 

[tmcw]

Many thanks for the pointers Ken.

I probed all the RAM chips, and pin 2 pulsed on them all. I probed all pins and got the following, with 3 exceptions (P=pulsing):

Pin
1 -
2 P
3 Hi, then pulsing during rug wipe
4 P
5 P
6 P
7 P
8 Hi
9 Hi
10 P
11 P
12 P
13 P
14 P
15 P
16 Lo

The 3 exceptions were pin 14 on chips 2-5, 1-3 and 1-4, which were all stuck low.

=============================
I also checked the other 2 chips you mentioned, and got the following:

1H:
1 Hi
2 P
3 Lo
4 Lo
5 P
6 P
7 P
8 Lo
9 P
10 Lo
11 P
12 P
13 P
14 P
15 P
16 P
17 P
18 P
19 P
20 Lo

4E:
1 Lo
2 Hi
3 P
4 P
5 P
6 P
7 P
8 Lo
9 P
10 P
11 P
12 P
13 P
14 P
15 Lo
16 Hi
====================================

Anything bad or unusual there?

 

[tmcw]

Had another bit of time, and had another look at the RAM chips. I swapped out the 3 that were showing no pulsing on pin 14, with 3 other chips which behave like the rest of them, and the machine seems a bit better.

Here's what I'm getting now:

Rug pattern is now clean, it looks normal (before, it looked like there was corruption, probably down to the bad RAM chips, as it seemed to improve as I changed each of the 3 with good ones).

On start-up, the ROM board flashes 1-1-1 twice, then goes blank.

The rug pattern then disappears, and I see the message "Initial tests indicate: RAM error 11".

After that message disappears I get one of two things:

1: On a reset (push-button), or quick power off/power on, the screen almost goes completely black, except for about 7 or 8 clusters of 1 to 3 pixels about the screen.

2: On a cold boot, I get a message about bookkeeping settings/high score being reset (there is a little corruption on the top line of text), I press advance, and I get the 7 or 8 clusters of pixels described above, but the text remains.

In both cases above, triggering the coin switch, I hear a noise which I think is correct for the insert coin sound, then another sound when I press the 1 player button, which isn't correct. But I hear no more sounds after this, and the screen doesn't change from described above.

Also, after the "Initial tests..." screen disappears, the ROM board indicator displays 0.

Changing around the RAM chips doesn't move the 11 error.

Any suggestions on how to proceed from here?

 

[tmcw]

Changed out a few more RAM chips, and got to a point where there is no corruption on any of the text during boot-up. Still getting the 1-1 RAM error though. I'm fairly certain that the RAM chip there is good, and I replaced the socket, so at a bit of a loss what to check or replace next.

Also, finally got a coin-battery holder in, and it seems to be functioning fine. When I start it up now, I don't get the message about high scores and bookkeeping data being restored, etc.

One thing I've noticed when probing the RAM chips is that all pins that strobe, strobe at seemingly the same rate, EXCEPT for pins 14 on RAM chips 3-7 and 2-7, which strobe at a very high rate. I've moved chips about, and it's definitely the location, not the chip that's exhibiting this. Is this significant, or normal?

I'm also wondering what the normal outcome of a RAM error is? Does the ROM board normally display "0" even with a RAM error?

 

[Dokert]

I'm also wondering what the normal outcome of a RAM error is? Does the ROM board normally display "0" even with a RAM error?

Sometimes it will display a "0" with a RAM error.

1. Check RAM 11 for broken traces to the socket.
2. Verify the voltages on the corners of RAM 11 also.

 

[tmcw]

Had a look at the continuity from the chip, through the socket, to the board, and out to the chips that they're directly connected to, and they all check out ok.

Voltage at pin 9 is 4.96V, same as with all the other RAM chips.

I started through the troubleshooter again, since the game seems to be getting a little further, and before where I was getting no pulsing on a few pins on 2D and 2G, they are pulsing fine now.

However, when I get to pin 19 on 1G, it's still stuck low. Troubleshooter says to "Check MPU Data Buffers" if no pulsing (pins 2 to 9 on 1G are ok). How do I check the MPU data buffers?

1G equivalent on a Defender board is 1K, if I understand the schematics correctly, a 74LS245 octal bus transceiver.

 

[YellowDog]

1G is a 74LS245 on the Stargate board as well. You can get the Stargate (as well as a lot more) schematics at arcarc.xmission.com.

Pin 19 is the not Enable line, so if it is low, that is a 0. So not 0 is 1 or enabled. So Pin 19 should be good. If it wasn't, I think you would be seeing all kinds of issues. Pin 19 is tied to the BA signal on the CPU (pin 6). According to the MC6809 data sheet the BA & BS lines should both be zero when the CPU is running normally.

IIRC, 1D, 2D, 2G are the MPU data buffers.

ken

 

[tmcw]

Yes, I have downloaded those documents and printed them out. Working through the cpu logic check in the Stargate/Robotron document.

So if pin 19 isn't supposed to pulse, then that chip is ok. The 1D, 2D and 2G chips were tested in the previous step on the flowchart, and all the listed pins to check pulse ok for those 3 chips.

So I continued on through the flowchart, and everything was pulsing until I got to chip 6G (7432 QUAD 2-INPUT OR), where pin 6 was stuck high (pins 8 and 11 were pulsing). It says here if no pulsing to "Check Video RAM Control Circuit".

Does it mean this chip is bad? And possibly why I'm seeing the 11 RAM error?

I went down to the next few steps on the flowchart, and found pin 8 on 5K stuck low, and pin 9 on 4F stuck low, but I figure these could be dependant on the 6G being good?