Williams Repair Log by Mike Ranger
Well, I ve been fixing some Robotrons for Dave, and some of them really have me stumped. This has lead me to pouring over the schematics time and time again, and guess what? Some of it actually sunk in. So my intent here is to talk alittle about the various sections of the schematics and what certain chips do. With luck this may help others to narrow down problems.
A quick note...I have used Sean Riddle's page extensivley for my multigame, as he talks alot about S/W and addressing, but specifically on a slant towards his well done multigame. I wish to talk here more about what the different chips do, to help others in problem solving.
Copyright: At this point, I offer this text in its entirety, to all who may make use of it. Please quote the author if it is reproduced.
Disclamer: These are my Ramblings. If you use the information ,I take no responsibilty in it accuracy, and will not be held resposible for any damage caused by its use! If you find an error (and I'm sure there are some) tell me!
Section 1 - Memory Errors
Got a 1-3-1 error? For Gods sake don't start throwing out chips (mail them to me ;-) ). When the ram is checked via self test, the first chip done is 3-1. If this chip is reported bad, it is almos always a power related problem. Chckt the voltages right on the pins of the RAM.
1 - -5 v
8 - +12 v
9 - +5 v
16 - Ground
If there is a missing voltage, shut her down. Leaving the power on too long with a missing -5 can damage these pesky RAM chips. Check the supply, moreover, check the connector to the main board and the solder job on the underside of the board. The joints have a habit of cracking.
To move on, the self test does 3-1, 3-2, ...3-8, 1-1, 1-2, ...1-8, 2-1, 2-2, ...2-8. So if a memory location fails, swap the suspect chip to any location already tested. If the problem moves with the chip, replace it. If it stayes in that location, you may have a bigger problem. I cannot suggest exactly what to look for, but read on to the point where I talk about memory access.
Section 2 - Memory Addressing...
There are two potential users of memory on the system. The CPU and the Video Output. The btw correlates to having 2 video decoder ROMS. Lets talk first about the CPU.
If the CPU wants to get at the RAM , it uses A8 to A15 and A0 to A7. The higher bits are first put through the video decoder ROM at 3G to produce a column address select byte (CAS) and the lower bits are used as row address select bits (RAS). Actually since the RAM only needs 7 lines, A0 is used at the bit 7 (MA6) of the CAS, and only D0 - D5 are used for CAS out of the decoder ROM.
The 74153's are toggled to select one of 4 output scenarios. CPU address lines RAS and CAS, or video address lines to the CAS and RAS. Now these address lines are presneted to every ram chip, and every ram chip then outputs its data on to its own little bus. If the CPU wants the data, it will enable the 74LS374's corresponding to the bank it wants and the data is then put onto the main data bus. Alternatively, the CPU can select the write enable lines (so can the blitter chips) to put data from the main data bus into the RAM chips for a particular bank. As you may or may not be able to see thare are many possible points of failure in this loop. History has shown that the '374's will fail from time to time causing the CPU to report bad memory chips.
The video output is a bit different however...There are 4 binary counters constanly couting up, then getting reset. The output of this count drive, among other things, address lines gor the memory (RAS) and the decoder at 3C (CAS). When the '153's are passing this data through, the memory output goes down to 8 74166's. One bank of 4 is for normal mode, the other is for cocktail flip screen mode. These chips are clocked at 6 MHZ and simply shift the data out in 4 bit blocks to the 74LS257 (which appears to only select which bank of 4 166's to select from for flip screen). This data is then fed to the 74LS257 at 2C. 2C is the chip responsible for feeding the 7489's either address lines od the RAM data. The 7489's are the colour look up RAMS, which take the 4 bit info from the '257 and produces 8 bits of colour (two banks 7489's) and that 8 bits of colour ar buffered to the transistors which control the intensity of the RGB output. Alternativly, the 2C chip can send address data to the 7489, in conjuction with the CPU setting up the data to write to the RAM. In this manner the CPU can dynamically alter the colour pallet, choosing from 256 colours to get a max of 16 colours out on the screen at a time.
Section 3 Sync...
It is well documented that the Williams games have pin 7 ov the video output
as Composite Sync (positive). Too bad williams did not document this.
Another documented fact (Joust to Jamma) is that in order to produce a negative
comp sync, you simply (?) add an inverter to this. I think I have found
an easier way.
On the schematics, there is a zero ohm resistor called W1 on the input to the 4 XOR gates ar 4A. This is incorrect. W1 is actually a jumper off of 6A. The incorrect W1 is actually a trace on the board under resistor R34. This trace, when intact has the function of making R34 useless. It has one end on ground (via the trace) and the other tied at 5 volts. The trace then holds the second input to 3 of the four XOR gates low, and by definition, the XOR gates are reduced to a buffer. These gates are Comp sync, H Sync and V Sync. If we now cut the trace, R34 is now a pull up resistor , making the second input to the 3 XOR gates high, and again, by definition makes the XOR gates into NOT gates. Voila - negative Comp sync. Works Great.
Williams (Joust, Robotron) 3 different sets
Symptoms: No sings of life, power fine.
Fix: Using both a scope and a logic probe, looked for activity on the
CPU. Found the E (external clock) line to have no activity. Looked
at the 7474 at 7F. Pin 8 was toggling (/Q) but pin 7 (Q) was not.
Replace the chip and the board roared to life.
Symptoms: Ram 1-1-1 error, known good ram chip.
Fix: Verified the ram chip in position 3-1, it was good. Verified the
socket be scoping out the address lines with 31 (common address) as well as
the output to 1H (74374). Replaced the 74374 at 1H and 4H (in case bank
3 was interfering). No change. Realizing that it would be difficult
to scope out a problem with addressing, I replaced all address multipexers (4E,3D,3E,4F).
Still 1-1-1. Tried to verify the write enable lines, but again, difficult
to pinpoint a problem without a logic analyzer. Swapped out 4M and 4L
(7410's) which do the write enables. Problem is resolved. Swapped
them back in one at a time and the problem was with 4L and the /WE1L.
Symptom: Every second line of raster, will not go into demo mode.
Fix: After an exhaustive amount of probing on the board around the syncs,
I finally found a bad 7400 in the blanking circuit. I used my logic probe
and found that both inputs (12,13) to the NAND gate were moving together.
I checked 6C (D flip flop) and found that the output /Q pin 8 was not responding
according to the inputs. Replaced it, the problem persisted. Replace
4B and the raster was restored.
When looking for the problem with the "Demo Mode", I found that the CPU was not letting the blitter chips (on the known good rom board) get bus access, as well A11 was not toggling much at all. I get ready to replace a bunch of chips, but on a whim I decided to test the CPU in another machine. Indeed the problem moved with the CPU. The CPU is now in the garbage.
Willaims Bubbles:
Symptom: Ram 1-3-1
Fix: 12volt reg reading 5 volts on load, 12 volt off load.
Replaced the 3055 transistor on the heatsink assembly.
Williams Robotron:
Symptom: No activity on the screen or LED Diagnostics.
Fix: Watchdog constanly resetting the CPU.
Found a bad 7404 at 5A which, among other things, provides a clock signal
to the counter in the watchdog circuit.
Williams Robotron:
Symptom: Ram errors, battery damage
Fix: cleaned the board with a 50% vinigar
solution and a toothbrush and rinsed. Board roared to life
Williams Stargate:
Symptoms: Bad 5 volts on most chips
Fix: Found a chip from a previous repair installed
backwards. Replaced. Now memory errors. Found a bad decoder
4, by tracing back after finding the Write enable for the chips were not
toggling at all. Replaced. Now CMOS Ram errors. Replaced
the CMOS RAM and started playing!
Williams Stargate:
Symptoms: Misc Ram errors
Fix: No obvious problems, but the write enables
coming from 4L and 4M did not look right (scope) (gut feel). Replaced
them both, then had to replace a RAM socket and it Roared to life.
Willaims Linear supply
Symptoms: 12 Volt LED was not lit.
Fix: The 12 volt level was there, turned out to
be a bad LED!!!! Replaced.
Williams Linear suppply
Symptoms: No 5 volt, or -5 volt output.
Fix: One fuse holder was corroded, cleaned it up and
I got -5 volts again. On a whim, I replace the regulator for the +5 Volts
(it was socketed) and the Power supply is now fine.