19th December 2019
CAPS0ff has been busy decapping some of my Semicom 87C52 MCUs and as a result the Decapping Page was updated several times over the
last few days. If this continues, there should be at least 3 more 87C52 MCUs decapped relatively soon-ish :-)
Please support CAPS0ff with funds if you can so they can work on some of the more difficult decapping jobs that are waiting in line.
10th October 2019
Continuing with the Commando repair, there were no additional issues on the bottom board and the game was fully playable after swapping out
those 5 bad logic chips I mentioned previously. The only remaining issue is no sound. There are no Fujitsu logic chips on the top board now
so it can't be that. The Z80 has the clock and reset but there is no activity on several pins of the Z80. Nor is there activity on the ROM
or RAM. The ROM was read and checked good against the MAME ROM set. The RAM was the next suspect. I swapped out the RAM and it didn't help,
but I was not able to test the RAM. Luckily I added a socket for the RAM so I tried another RAM and it still didn't work but it was slightly
different and started making some random noise. I went through about 10 different 2016/6116 RAMs and then eventually the Z80 came alive and
became active. Strange that so many RAMs did not work, but maybe this board is fussy about the brand of RAM or something. Coining up showed
additional activity on the Z80 so the sound program was now running but there was still no sound output! I touched the amp with my finger
and it made the normal 'pop' sound so the amp was ok. I also touched the input pins of the 2x YM3014 and LM324 chips with a wire connected
to ground and the speaker also popped, meaning the YM3014 and LM324 chips were also ok. I checked the output of the YM2203 chips on pins 22
and 23 and didn't see anything on my logic probe. This is normal for digital sound chips. I probed the same pins with my oscilloscope and
there was no activity. Then I tried a different working game that uses a YM2203 chip and there was activity on the oscilloscope on pin 22
and 23. So now I knew what a working chip looked like on the scope, I went back and re-checked the chip on the Commando board and
it was definitely not outputting anything. I decided to piggy-back the YM2203 and to my surprise I heard sound. I was very scratchy and
noisy but definitely sounded like the sounds from Commando. I pulled the chip and swapped over a working YM2203 chip and sound was fully
restored.
And here's a quick video of the game running and me (badly) playing it sideways on my test rig....
Update 11th May 2022
The sound died again. Sometimes it would play part of the teletype sound at the start but it would get corrupted and then stop playing or
sometimes there was no sound at all. It turned out to be the other YM2203 that died. I first removed the top YM2203 and tested the game and
the sound worked perfectly, but of course was missing a sound channel. I swapped in a known good chip and the sound was fully restored. I
also tried different sound program RAMs and now they all work. So the strange behaviour with being fussy about the sound RAM was caused by
the YM2203 chips. The data lines on the YM2203 are tied directly to the CPU so a faulty YM can bring down the whole sound system.
5th October 2019
Getting back to the Commando repair, while diagnosing the rest of the faults the entire thing just died and now displays nothing on screen.
There is no clock on the CPU. If you check the CPU page on the schematics in the previous repair log, you will see this clock comes from the
74LS367 at 8K (Ø MAIN) then to page 6/8 (Characters, also shown in previous log) through a 74LS32 at 5H which was checked as good in the last repair
log. The input on pin 4 of the 74LS32 is shown as ØB which comes from page 5/8 titled 'Synchronous' which is shown below....
The ØB signal is coming from a 74LS367 pin 9 at 3K. The input of that is pin 10 which goes to a 74LS74 at 2M and then to a 74LS04 at 2N. Look
closely and you can see the 12MHz crystal is connected to the 74LS04 and of course all of the outputs are dead. This is another bad Fujitsu
chip which needs to be replaced.
Changing it has brought back the clock on the CPU, but the board is still dead!
On the same schematic page there are 2x 74LS74 and 1x 74LS08 chips which are Fujitsu branded. I checked the inputs and outputs and they are
ok, but I changed them just to get rid of them. There was no change on screen.
At this stage it gets more difficult because the PCB is slightly different on the bottom right corner of the CPU board and that part is specific to the bootleg board so I don't have a schematic. Let's have a look at the chips there and figure it out....
In the bottom corner, the 74LS245 is connected to the CPU and the PAL. On the 74LS245, pin 19 is OE (Output Enable) and is active low. This
pin is high so the 74LS245 is not being enabled. At this stage I could trace out the PAL to LS245 wiring, but we have to be realistic about
spending an unknown amount of time chasing a rabbit down a hole. I have already changed many of the Fujitsu chips on the top board so more
than likely the fault is on the bottom board. I decided to swap out the lower board with another board I have here that is also faulty. To
my surprise the game now comes up! It is of course still not working 100% and has worse graphical faults than before but it does show
something on the screen so it looks like one or more Fujitsu chips that were working have now completely failed. I put the other faulty
bottom board back. Now I could go along and test the inputs and outputs of the same Fujitsu chips already identified as possibly faulty, but
I will do this a slightly different way. One technique to find a fault is to piggy-back a chip on top of a suspect chip and note if there is
any difference to the screen or the output pins on the chip. Whether it works or not depends on how the chip has failed. If it failed open
the piggy-back will usually improve it. If it failed shorted then the piggy-back usually won't help. I plugged a 74LS32 on top of the chip
at 6B and the game fired up again, back to the same faults as before. So the Fujitsu-branded 74LS32 at 6B just died.
Now I will show a technique to remove a chip without using expensive equipment like a powered desoldering gun. It is as simple as
snipping off the legs of the known faulty chip with fine-pointed side-cutters, then using a soldering iron and tweezers to remove the pins
one by one then sucking out the solder to clear the holes. Using the side-cutters, snip the legs at the top near the plastic package, and
remember NEVER do this to a chip that you think *might* be faulty if it is a custom chip, because snipping legs off a chip
destroys it beyond repair. Only do this when you know a chip is bad and it's a common off-the-shelf chip.
Here's some pics showing the tools used, some of the legs cut, then the plastic chip removed, the legs removed one by one with soldering iron
and tweezers and the holes cleared out with a manual solder sucker (i.e. the ghetto way to remove a bad chip). Notice how the PCB
looks like new because none of the solder-mask was scratched like when using a powered desoldering gun. One tip to improve the solder sucker
is to spray some light oil (WD40 etc) inside the tube then the sucker will suck faster creating more suction and improving the efficiency of
the manual solder sucker.
The last pic shows the game working but all sprites are missing. Now you can see how having a multi-stack board with cables joined only on
one side is a good thing. To access any board just flip up the board :-)
At this stage there's no point going over the lower board and showing checking the same 74LS00/04/08/32 Fujitsu chips that have already been
identified as bad since that info was shown in the first repair log. I will do that myself and come back when it is done.
In the next repair session I will continue with the remaining sound-related faults and if anything unusual happens on the lower board chip
swap I will document it here.
However I will leave you with a very large pic of 5 chips piggy-backed (marked with a red dot) and the effect it has on-screen ;-)
These chips are all Fujitsu branded. Locations are....
74LS04 at 9A and 1F
74LS32 at 8C and 9C
74LS00 at 8B
12th August 2019
I recently dumped an alternative bootleg version of Commando (a classic game by Capcom released in 1985). It uses a mix of ROMs from the
original PCB and other bootlegs, but one ROM was different. It was added to MAME yesterday as 'commandob3'. The PCB was sitting around on a
pile of junk for about 15 years in a non-working state but is actually in pretty good condition so I thought it would be interesting to fix
it. It should be relatively easy for a few reasons... it has no custom chips, the boards are joined with a single cable (you will see why
this is important later) and also because it uses several logic chips made by Fujitsu with date codes that are known to have failed. For a
change, I will make this into a basic repair tutorial for someone wanting to look into repairing an arcade PCB and has no experience. But
with a difference. Rather than just say I scoped this and that and replaced this and that like other people do, we will look in some detail
at how the chips work and use some inexpensive tools to help diagnose it. That way you might actually learn something and be able to repair
your own PCB or at least pique your interest to do so later. This is probably going to be quite long so sit back with a hot or cold drink
and read on....
Here's a pic of the board.....
Yes this is a bootleg and generally considered to be not worth much, but it's in great condition. If you just want to play some classic 80's
games in an arcade cabinet this is what you want. Well, hehe!! Actually for classic 80's and 90's games an X-in-1 is what you want or a MAME
PC, but if you want something that resembles the original hardware, a bootleg is the best way to achieve it. For example I used to have an
original Gyruss PCB but sold it while it was still working and later got a non-working bootleg PCB for nothing and fixed it relatively
easily. I doubt the original board is working now because it has several custom chips and is not easy to diagnose. More modern boards can be
a nightmare to fix due to having sometimes dozens of custom chips that may or may not have failed and you'll be tearing out your hair trying
to figure out what's wrong with it and after many wasted hours you will just give up. Bootlegs like this one use only common logic chips,
EPROMs, RAM etc so can always be fixed no matter what goes wrong with it. This bootleg is kind of unique among bootlegs because it has a
chip layout that matches exactly with the original board and doesn't use any of the plug-in daughter boards often found on bootlegs (a
common failure point is having bad connections between these plug-in boards and the socket on the main board). Even better, the original
schematics are available so we can use it to help with the repair.
The tools you will need are a soldering iron, solder, flux, side cutters, multi-meter and a logic probe (with 3 LEDs; low(green), high(red), memory/pulse/clock(yellow), and a piezo
speaker.... all of these features are required and important!). All of this stuff is available on eBay etc and this is the bare minimum if
you want to be successful at any PCB repairs. You can't fix anything electronic without these basic tools and I always laugh when I see
people on forums ask for help with an electronic problem but don't even have the basic diagnostic tools LOL! Well you may be able to do it
without the logic probe if the PCB is not a little computer like an arcade game, but as far as arcade games are concerned it will be much
more difficult without some way to 'see' the status of a chip and this is what the logic probe provides. You don't need an oscilloscope, and
in fact most people who think they need one to repair stuff don't really know how to use them anyway. We will use
the logic probe as the main diagnostic tool, together with the most important tool, our brain. Oh and a few senses too; sight, hearing and
touch... but not the other two. Sorry to disappoint but no, we are not going to be sniffing or licking any PCBs ;-)
When powering on the game it only shows a black screen. The first thing to do is check the clock and reset on the CPU. Commando uses a Z80
as the main CPU. Refer to the diagram below....
On a Z80, clock is pin 6 and reset is pin 26. Probing the clock with the logic probe shows that the yellow LED is flashing constantly. This
is the clock, so that means the 12MHz master crystal is OK and the logic that divides it is 'probably' OK. We have some kind of clock so we
are happy. To know the frequency you need a frequency counter or oscilloscope but let's assume the clock is OK. To check the reset, have the
PCB powered off, then position the logic probe on pin 26, then power on and observe what happens. The pin should be low for about 1/4 second
then go high. It does so the reset is OK. Some early CPU resets start high and go low, so be sure to check the datasheet for the proper
operation of the reset pin, but regardless there will be a clear transition from low to high or high to low. So while probing these pins I
noticed that the Z80 is very hot, almost too hot to touch. This probably happened when the game was left on for a long time in this non-working
state and high frequency rapid toggling inside the CPU destroyed it. I figured it is bad so I swapped out the CPU for a known good
chip. I also noticed a broken 10k resistor network near the main program ROMs so I changed that as well. When powering on it now shows
something on the screen....
At this stage I could just start changing out all the Fujitsu logic chips, but that would be too easy and you will learn nothing. In the
voice of Siegfried from 'Get Smart'... zis ist KAOS, zeir vill be no shotgunning here!
First let's have a look at the CPU page on the schematics.....
Thankfully this was scanned at 600DPI so it very readable. For the purposes of loading time the image has been reduced but the original pdf
is very nice and the pages are much larger. Click the thumbnail and then click the 'four-arrows' icon in the bottom corner of the image to
view it full size.
The circuit starts at the top left corner. The reset 'high' comes from the PST518. On a bootleg this is just a diode, a 4.7uF electrolytic cap and some resistors. The
signal travels through a cap which charges up then goes high after 1/4 second then to a 74LS14 at 11J and then to the Z80 pin 26. On the
right side of the Z80 there is a 74LS138 at 9J and then a 74LS08 at 9L. Looking on the PCB it can be seen that the 74LS14 and 74LS138 are
not Fujitsu chips but the 74LS08 at 9L is a Fujitsu chip! Looking at the rest of the PCB it can be seen that all of the LS00, LS04, LS08 and
LS32 chips are Fujitsu chips with date code '8502' or '8504' and there is a very high chance that all of them are bad. There are also a few
other random Fujitsu logic chips scattered across the board such as 74LS74 but these are not all exclusively made by Fujitsu and in any case
were made in 1983. I might look at those later. I think this Fujitsu issue affects most logic chips made between 1984 and 1985 at the very
least. These Fujitsu logic chips are rampant on Konami games from the mid 80's as well, in this case especially LS07 and LS253. But I have
seen failed Fujitsu chips as late as 1992 on (for example) Lethal Enforcers. So basically you can't trust any Fujitsu logic chip no matter
when it was made. I suppose that's why eventually they stopped making logic chips hehe! Thankfully these are all simple logic chips so easy
to check in-circuit. Let's start by looking at the LS08 at 9L. I have seen videos on YouTube where people probe logic chips with an
oscilloscope or logic probe on an old Commodore computer (C64 etc) and then announce that it looks OK. You can't just probe a logic chip and
know if it is good or bad, you must know what is expected on the outputs given the current inputs including which pins are inputs and which
pins are outputs, and to know that you must look at the datasheet. I like the old school 80's Texas Instruments datasheets because generally
everything I need to know is shown on the first page in a very clear and concise manner. One tip, if you plan to do a lot of logic repairs,
get hold of a real paper TI Logic Databook because it's a lot faster and nicer to look at real paper than looking at the pdf datasheet,
especially if the repair area is not near a PC. You can also access datasheets on a PC of course, or a tablet or phone. You can also
download the entire TI Logic Databook in PDF form and keep it for reference but be sure to get an older Databook as some info has been
removed from the first page for some reason. The older 80's Databooks are much nicer. Here's the datasheet for the LS08....
This is a Quadruple 2-Input Positive AND Gate. It sounds complicated but basically it just takes some inputs and ANDs them together and
outputs the result. This is known as 'Boolean Logic'. Some logic chips can be VERY complicated but this chip is very simple.
Referring to the pinout, there are four individual and separate sets of two inputs (A , B) and one output (Y). 1A & 1B pairs with 1Y, 2A &
2B pairs with 2Y and so on. The logic diagram on the datasheet shows how it looks when drawn on a schematic and if you look at the Commando
schematic you will see the same symbol for the LS08 at 9L (well it's nearly the same, there are variations on how symbols are drawn but it's
the same thing). It is shown four times because each set of 3 pins (A,B,Y) is separate. The truth table on the datasheet shows how the
outputs react based on the low or high input levels. X means 'I Don't Care' so only one low is significant or two highs. If both of the A
and B inputs is high, the output is also high. If either A or B is low then the output is low. If there is a low and a high on A,B the high
is ignored (i.e. Don't Care) and the output is low. In Boolean terms, both pins must be 1 to output a logical 1 otherwise you get a 0.
That's all there is to it for this chip. The pins are grouped together and easy to remember so now we can test the chip in-circuit and see
what the logic probe tells us about the status of this chip. With the game powered on, probing pins 1 and 2 shows nothing! The game may have
been trying to do something but has crashed instantly at power-up. If I probe the pins while powering it on, pin 1 shows low for a second
and pin 2 shows nothing. The output pin 3 (Y) shows nothing also! If the chip was working, the output should be low because pin 1 is low.
This chip connects to the ROMs CE pin (active low chip enable) so the ROMs are not being enabled and therefore the program can't run.
Probing the remaining pins on the chip shows similar inputs (high or low) and the output pins are all nothing, so all the outputs are dead,
meaning this 74LS08 at 9L is toasted and must go!
We have now positively identified one bad chip. Yes, four pages of text and one logic chip changed... well I did say it was going to be in-depth
hehe! At the top I mentioned this tutorial might be long but that might have been a massive understatement. It will get more
interesting very soon so don't get discouraged ;-)
So what does the game do now? It shows even less on the screen just some blocks of garbage and a black screen. Urgghh!
BUT!!!! It seems to be running?!?!!?!!! It is cycling between three screens that mildly resemble game screens shown in the attract mode.
When I coin up and press start it shows the in-game screen! The first black screen was probably the 'For Use In Japan' message. Wow it's
running! Well it's kind of running, perhaps more accurately it is showing signs of life but needs emergency intensive care!
At this stage I like to probe the address and data pins on the CPU and as expected there is a lot of rapid activity on the logic probe. As
the attract mode screens cycle through, the activity also changes on the data pins. This means the main program is running! For now I will
leave the CPU section alone.
The backgrounds appear to be mostly OK (with flickering) but there is no text on the screen so let's have another look at the
schematics. This is the page that deals with characters....
Checking the schematic we can see there are four chips that can possibly be bad, based on the fact they are Fujitsu chips.
These are 74LS00 at 3F, 74LS08 at 6F, 74LS32 at 5H and 74LS04 at 3E. There's no order to checking the chips but since we already know how to
check an LS08, while I have this in my memory I check it and the logic probe reveals pulsing inputs and pulsing outputs but only on pins 11,
12 and 13 so the chip is working at the moment. Most of the other pins including those that connect to the LS74 at 4H and the LS04 at 3E
are doing nothing and the outputs are low so this chip isn't dead. Since this is about not only fixing it but totally obliterating all signs
of Fujitsu chips from this board, I just removed the chip for good measure and swapped in a non-Fujitsu chip but as expected there was no
change on screen. However this has eliminated a future fault at 6F, which as you will see later can be significant in the repair
process.
The next chip I am going to check is the 74LS32 at 5H and here's the datasheet....
This chip works in a very similar way to the LS08 but instead of AND'ing it is OR'ing. Referring to the pinout, as with the previous LS08
analysis, there are four individual and separate sets of two inputs (A , B) and one output (Y). 1A & 1B pairs with 1Y, 2A & 2B pairs with 2Y
and so on. The truth table on the datasheet shows how the outputs react based on the low or high input levels. This time only one high is
significant or two lows. If either of the A or B inputs is high, the output is also high. If both A and B are low then the output is low. If
there is a high and low on A,B the low is ignored (i.e. Don't Care) and the output is high. This is another very simple chip. The pins are
grouped together with the same style as the LS08 and also easy to remember so now we can test the chip in-circuit with the logic probe. With
the game powered on, probing pin 1 gives nothing and pin 2 gives nothing. Pin 3 (Y output) shows high. This doesn't really tell us anything
other than maybe the outputs are not dead (but they can be stuck high hehe!). Pins 1 and 2 are missing for some reason. Probably some other
Fujitsu chip triggers some other chip that provides those inputs and that chip is not outputting anything so the inputs on the LS32 are not
present. Also pins 1, 2 & 3 are not used in the character circuit so not yet relevant. But checking the other pins shows similar results,
either no inputs or just a clock signal (flashing yellow LED). But at least we now know how this chip works. Let's move on.
The next chip I will check is the 74LS04 at 3E. Check the schematic above again and you will see this connects to the LS08 at 6F, which
connects to a LS74 at 4H and then to the LS32 at 5H that we just checked! Now it's beginning to make sense. Obviously one of these connected
chips is screwing things up.
Here's the datasheet for the LS04....
This is even simpler than the other two chips we have looked at so far. This chip is a HEX Inverter. Hex meaning 6 of them in this chip and
inverter meaning it takes the input signal and outputs the opposite signal. It is basically just two pins for each circuit, in and out X6.
This time we will move in for the kill and just check the pin on the schematic in the character section. When probing pin 5 it shows a high,
but occasionally the yellow LED is pulsing so a signal is coming into the chip. The output is on pin 6, but probing it shows nothing!
Checking some of the other input pins shows activity but all of the outputs are dead! I swapped out the chip but it did not change anything
on screen... urgghh! But now all the outputs are working, with a low input coming out high and a high input coming out low, so for sure
this chip was bad.
There's one chip remaining in this section that hasn't been checked yet, the LS00 at 3F. If you look at the pic above you will see our not
so friendly Fujitsu 74LS00 chip right next to the LS04 we just changed.
Here's the LS00 datasheet....
The LS00 is a Quadruple 2-Input Positive NAND Gate. The pin layout is the same as the LS08/LS32 and it works exactly the opposite way as the
LS08. With this chip, if both A & B are high the output is low. If either A or B is low then the output is high. In Boolean terms, if both
inputs are 1 it outputs a logical 0, otherwise you get an output of 1.
Referring to the schematic you can see that the LS74 at 4H connects to the inputs of the LS00 at 3F and the output pin 3 is tied to the
character RAM at 7F to the WE pin (write enable). Probing the input pin 1 & 2 shows activity. Pin 1 has all three LEDs on (low+high+clock).
As the three attract mode screens change, the activity changes slightly, then goes back to the regular pulsing pattern before. Pin 2 shows a
low and occasionally the yellow LED pulses. Probing pin 3 shows nothing!!! Probing the other pins shows activity on the inputs and nothing
on the outputs so this chip is dead. I changed the chip and now it's starting to look like Commando!
The screens still show garbage blocks which I'm assuming are partial sprites. In-game there are no real changes from the previous in-game
pic other than the text is now visible (score etc). Notice in the 2nd pic above that the top board is folded back like a page in a book.
This board has a cable attaching all the boards but only on one side so it makes it easy to work on the individual boards while it is
powered up. With a lot of newer multi-board games, the individual boards plug into special connectors obscuring the lower boards and the
game will simply do nothing if everything is not connected. This makes it impossible to work on the board while it is running. I didn't get
a shot of it, but the start-up text 'For Use In Japan' now shows (with garbage blocks). I then went back and checked the LS32 at 5H.
Checking pin 4 shows a low with a clock pulse (solid green LED + yellow LED flashing). Pin 5 has a similar signal and pin 6 (Y output) also
has a similar signal so the inputs are being correctly OR'd together and output. Checking pins 13, 12 and 11 shows similar results. Pins 10
and 9 are low but for a few seconds pin 9 shows low+clock pulsing, then back to low. The output pin 8 also shows the same low and then
low+clock for a few seconds. As we can now see, there's something happening with the text on the screen when the yellow LED is flashing. To
test that, I powered on and I get the black screen which 'types' the 'For Use In Japan' characters on screen, and pin 8 has the same pulsing
low+clock which is in sync with the text showing on the screen. This happens when it shows the start-up text or when 'Insert Coin' is
flashing. When no text is being manipulated on screen the output pin 8 stays low. We have proven that the 74LS32 at 5H is OK. As before, I
removed this chip anyway and put in a non-Fujitsu chip to avoid a future failure.
We have made significant progress and you should have learned something along the way. This is getting very long so I will leave this
tutorial here for now and continue later.
28th July 2019
Update: It broke again hehe! Small fix at the bottom of this section.
Here's another Neogeo repair. This time we have a NEO-MVH MV1 older single slot board.
The battery has leaked but there's virtually no PCB damage. Actually on this board the battery is relatively far away from the rest of the
parts that even moderately severe leaking won't effect the operation of the game and the game will work fine without the battery. I always
laugh when I see someone ask for help with a repair on neogeo forums and there are comments from clueless people asking if the battery has
leaked then they parrot the usual things to check around the battery area and pull other solutions out of their ass to make themselves look
important LOL! For the purpose of this repair the battery leakage is extremely minor so let's assume the battery has not affected anything
important ^_^
The board was initially only showing garbage on screen and resetting. I had previously worked on this and found a corroded trace between the two work RAMs.
I patched it with a wire on the bottom side and now on power-up it shows an error about color RAM and when using the Neogeo diagnostic ROM
it shows basically the same fault. Note that neither of those RAMs were anywhere near the battery...
The color RAM error is very unusual because there appears to be only a 1-bit error (write 5555, read 5554). I pulled and tested the two
color RAMs as well as the three logic chips nearby.
They all tested good so I used a multimeter to beep out the connections between those RAMs and the other connected parts. The RAM is mostly
connected to two custom chips NEO-G0 and PRO-B0, a few pins on the 68000 and some adjacent logic chips. There were no connection issues. The
soldering on the custom chip PRO-B0 looked in poor condition so I re-flowed the legs and re-tested but it did not affect the issue. I went
across each of the legs on the RAMs with a logic probe looking for irregular activity and then compared it with another working board I have
here. I found that all the pins on the RAMs were the same except one, pin 2. Here's the relevant section on the schematics....
Unfortunately and as usual, this schematic was scanned by someone who was clueless to what a schematic is used for, resulting in most of the
text being very difficult to read and in some cases totally unreadable. The actual schematic is from a MV-1F board but it's mostly the same.
Luckily I have a working PCB to check the connections. The RAM pin 2 is a signal called PALBNK (hard to read) and it goes off to a 74HC259
8-Bit Addressable Latch logic chip that is on the other side of the board near the 60 pin connectors. That chip is shown on the above schem
at the top left of the page (and yes, the actual chip 74HCxxx number is totally unreadable!). The way this chip works in the simplest form is it accepts some address
inputs on pins 1,2,3, active low enable/clear inputs on pins 14,15 and a data input on pin 13 and outputs on the other pins depending on how those
inputs are set (high/low etc). The 74xx259 data sheet has a truth table which can be followed but in this case it is not necessary to study it
because the inputs come from 2 custom chips (NEO-IO and NEO-E0). When I probed the RAM pin 2 there is no activity, but on the working board
pin 2 has some activity for about 1/4 second when powering on using the diagnostic ROM. This signal comes from an output on the 74HC259 pin
12, shown on the schematics as the PALBNK signal. I pulled the 74HC259 and it passed the test on my logic IC tester so the problem has to be
one of the input signals, meaning a possibly suspect custom chip.
Just for curiosity I plugged in the Unibios ROM which has been modified to skip most of the start-up checks and to my surprise the board
boots up. It shows the Unibios menu which is normally accessed by holding buttons A+B+C on power-up (allows changing region etc). This tells
me that the buttons are stuck on, very similar to the previous MV-1F repair I did in June. I checked the resistor network packs in-circuit.
With one multimeter probe on pin 1, the resistance on the other pins should read 1k, 1.2k, 1k, 1.2k, 1k, 1.2k, 1k, 1.2k & open (or whatever
the normal resistance is on the PCB between VCC and GND if checking in-circuit). On a different board I've seen 1.3k measured too, but
those are also OK. I found a couple that had the wrong resistance on the pins so I pulled a couple from a junk board and swapped them over and
now the board shows the normal test pattern on power-up without a cart plugged in. I plugged in a game cart and the game appears to work and
the colors look OK. I guess the 1-bit color RAM error is so small that it has no visual effect on the game. This is an unusual case where a
solution is not a full fix but is enough to get around the error. In a case like this it's not worth troubleshooting further so I'm calling
this issue fixed and the Unibios ROM will stay on the PCB permanently.
So I coined up and played the game. Controls are OK but there's no sound. If I listen carefully with the volume up full I can *just* hear
the sound, or at least some kind of sound but it's not correct. I thought maybe the YM3016 DAC might be bad so I pulled it and tested it on
a different PCB but it was OK. Then I looked closer at the pic I took and realised what the issue was LOL! There's some minor corrosion on
some vias that are connected between the YM2610 and the 68000 CPU and one of the vias measured very high resistance. I scratched away the
corrosion with a fiberglass pen and the connection from the via to the YM2610 went totally open! I patched the track and re-tested and now I
can just hear the sound and it's correct.
The volume is still very, very low and almost inaudible and there was a buzz coming out as well. Moving the volume pot did not change the
volume. Here's the relevant sound section on the schematics....
Referring to the schematics, it can be seen that the outputs from the 4x 4558 op-amps go to both the headphone section on the first page and
the speaker section on the second page. I plugged in a headphone and was amazed to hear that the sound is perfect and loud and adjusting the
headphone volume pot makes the volume louder or softer as expected. This means the problem is on the second page of the sound schematic. I
suspected that some of the electrolytic caps might be bad because they are all some shitty 'TK' brand (LOL!) so I pulled all of them
and tested them and as expected many were bad. All of the 470uF caps were totally out of specification and one cap even thought it was a
resistor LOL! I changed all the caps in the speaker section but it didn't fix the volume issue but it must have improved it slightly. Even
if I couldn't hear it, a cap acting as a resistor can't be a good thing for the audio quality ;-)
I also changed the main power AMP but again there was no change to the volume.
Strangely when I switched the mono/stereo switch from mono to stereo to buzz went away, but the volume was not louder. Using a multimeter I
beeped out the switch connections on my suspect board and on a working board and the results matched so the switch was OK. Referring back to
the schematic again, the only part that remains untouched is the volume pot! I measured the resistance of the pot and then moved it and
measured again and there was no change. I compared it with the working board and of course when the pot is moved the resistance changes too.
So it appeared that the pot was bad. This is not surprising really. Those pots are the cheaper type using a thin coating of some kind of
carbon-based material. If you check the specification of a standard 3-pin carbon-wiper pot you will see that they have a lifespan of 50 full
turns which isn't much! Now in this case, the volume pot is a custom part and the only place to get another is from another identical
board. However the headphone volume pot is the exact same 1kohm pot and being that it's the headphone pot has probably never been used. I pulled
both pots off the board and swapped them over and that restored the volume :-D
It isn't shown, but later I put the dead volume pot in the headphone pot position as it will never be used and won't affect anything. There
is a valuable lesson learned here. Don't randomly play with these special volume pots because they *will* go open eventually causing loss of
volume as I experienced. Then you will have to remove it and replace it with a pot taken from another board or swap over the headphone pot
to fix it. This also backs up the theory that the last part you change is always the one that fixes it, so to save time be sure to change the last part first! ;-)
Update: The sound broke again. It would work for a while then either the sound would go silent or it would play the wrong sounds. The test
mode sound test played the correct beeps so the Z80 communication was ok. It was another one of the vias connecting the YM2610 to the 68000,
the via just above the one I already patched. Between the track and the via there was some resistance instead of a direct connection. I
patched the track and sound is working again :-)
13th July 2019
Along with the trackball boards mentioned in the previous post, I've been busy making several other PCBs.....
The 2nd one off the production line is a cost reduced TC0070RGB. This one now costs $3 to build, complete with parts.
Most of the other PCBs are enhancements for Commodore computers....
- A3640/A3660 68040/68060 Accelerator card for Amiga computers (reversed by "Chucky". No changes, I just made some)
- Custom hard-card style XT-IDE to suit Amiga 1000 with A1060 XT Sidecar (in the style of the A2091 SCSI controller)
- Custom and enhanced Gotek drive containing all hardware modifications (OLED display, SD card, piezo, rotary dial etc) all built into the PCB and designed to directly
replace the floppy drive in an Amiga 500 using the existing FDD mounting points so it doesn't require a bracket or case modifications.
I will showcase all of these here later when I have built and tested them :-)
There are also 4 items for the Commodore 64....
- Another run of my joystick port switcher PCB (switch one joystick between port 1 and port 2 with the press of a button)
- Custom and enhanced Easy Flash 3 cart modified with a built-in dual ATMega1284-based SD2IEC with OLED display support. All on a cart PCB
the same size as the original. Will possibly be about the same price as the stock EasyFlash 3 to build plus of course it is even more
complicated to build than the original stock Easy Flash 3 cart. Again, it will be showcased here when I have built and tested it.
- PLA replacement chip which I call the "Open PLA", programmable for use with any of the computers that use it including C64, Plus/4, A1060 Sidecar, Commodore P600-series etc.
Again a cost-reduced design ($5) and using a proper 5V-rated part so no added B.S. to drop the voltage and no more having to pay $30-$50 for a C64 PLA ;-)
The 4th item is very special. It's an exact 1:1 reproduction of the Commodore 1581 3 1/2" floppy drive controller PCB that sits inside the case below the floppy drive.
The 1581 is unique among Commodore floppy drives for a few reasons. It is a 3 1/2" floppy drive for the C64. It came a bit late in the C64's life (around 1986) so it didn't catch on and is now difficult and expensive to buy. It also works with C128, Plus/4 and any
other Commodore computer that uses the IEC serial interface. It is also the only Commdodore floppy drive that is using all common off-the-shelf
parts other than a MOS8520 CIA.... also used in Amiga computers. The floppy drive is also a fairly common type used in Amiga
computers, but certain models of PC floppy drives will also work with small modifications.
This means unlike the older Commodore floppy drives, the 1581 can (in theory) last forever and be fixed no matter what is wrong with it :-)
This is not just a replacement PCB, it has the *exact* same look as the original too, including the quirky silkscreen, quirky footprints and
quirky PCB trace layout together with all the same texts and logos. Absolutely everything is in exactly the same place within 0.1mm.
Side by side the PCB is identical, other than being a better quality PCB. It's basically the exact same PCB design, just made 33 years later :-)
Here's a few pics of the first one I built. The first 2 pics show what the original looks like, the other pics are my 1:1 reproduction.
On first power-up it worked perfectly :-D
Here's the board installed in the original case..... it's a perfect fit :-D
6th July 2019
For those waiting, and for those who may be interested, I now have a new batch of Trackball I/O boards. This new board can control up to 3
trackballs and is designed to fit onto all the boards that use this interface..... System 16, System 18 and System 32, including of course Sonic The Hedgehog :-)
If you are interested, contact me.
11th June 2019
Something unusual just happened today. I fixed something purely by accident that I had no intention of fixing :-D
A local friend contacted me about a repair on his Neo Geo MV1C board. Someone had converted it to use a CR2032 coin battery and the battery
was being drained quickly. Obviously a poor decision because a CR2032 isn't up to the job. He wanted to replace the missing parts on the
bottom side of the board and put it back to the way it was originally and use a rechargeable battery or supercap. He asked if I could supply
the missing parts, which are all surface mounted.... one cap, one resistor and one diode.
These pics show how it looks now and what it should look like, and a simplified diagram of the circuit.....
A few years ago I bought a bulk lot of 6x MV1FZSB-2 boards.
I fixed three of them quickly as they just had RAM errors (Work RAM or Backup RAM), one had a broken track from the motherboard to the cart
slot (damaged via caused by corrosion) and the other two were worked on at the time but remained faulty and were put in my PCB storage area
for about 10 years. One of those is missing the upright cart slot PCB so that one will never work even if I fix the board.
Today I pulled out one of the faulty boards with the intention of taking the needed parts from the backup section to repair my friends
board. This board has a backup RAM error. I had previously replaced the RAM and checked all the connections. I found one bad via (again
caused by corrosion) and patched that track but it still had the backup RAM error.
The needed cap value is unknown so I started by removing the cap and testing it and found it measured 150nF. The resistor has markings on it
'221' which means it's a 220 Ohm resistor (for those who don't know how SMD resistor codes work, take the 22 and tack on 1 zero = 220). I have plenty
of caps and resistors in stock so I can just take those from my parts stock.
I removed one of the diodes and tested it on my little component tester and it said it was bad. I removed the other diode and it fell apart!
BTW, this little tester (model# MG328) is the best type to get because it uses a rechargeable 3.7v lithium-ion battery (type 14500) rather than the usual
9V batteries which are expensive and don't last very long. If you need to repair something with lots of transistors, resistors, caps etc,
simply remove each part one by one and test it with this nice thing. When the battery gets low just plug into a USB port on computer or USB
charger and the battery will charge up to 4.2V which is the full charge level of a 3.7V lithium-ion battery.
I didn't have the exact same diodes in stock but I figured a common 4148 would do the job near enough so I found some on a junk old laptop
motherboard in the same size surface mounted SOD-323 package (2mm x 1mm) and mounted it onto the MVS board. I powered up and now the board
is working. No backup RAM error :-)
So the backup RAM error was caused by the bad diodes! The battery is charging just fine as well so the diodes are working.
The 4148 diode is the same thing as the original 1SS352 (Silicon Epitaxial Planar Small Signal Fast Switching Diode). I don't know if the battery
will last as long as before but I don't care as long as it works.
Here's a pic showing the board with the original 1SS352 diodes replaced with 4148 diodes.....
I plugged in a cart and it coined up immediately, which is not correct. I went to the test mode and saw that coin 1 was stuck on which means
it is permanently tied to ground.....
On this board, coin goes from the JAMMA connector through a custom resistor network and then directly to a SNK custom chip 'NEO-FO'.
Hopefully the custom chip is ok and the problem is only a bad resistor network. I desoldered it and measured it. On the board pin 1 is
connected to VCC and pin 10 is connected to ground. On this suspect resistor network, measuring between pin 1 and each of the other pins (in
ohms) reads 1k, 1.2k, 1k, 1.2k, 1k, 1.2k, 1k, 1.2k and lastly pin 10 reads 1.2k. I desoldered another one from the 2-player control section
on the same board and measured it. Everything was the same except between pin 1 and pin 10 there was no connection, which is correct as
there should be no connection between VCC and GND. Bingo!
I soldered the good resistor network into the position where coin 1 connects and powered on. Now it doesn't coin up automatically. I tested
the coin-up and it works.
It isn't shown here, but later I found another correct and working resistor network on another junk MVS board and
soldered that on to replace the missing part and fully fix it. Problem solved :-)
4th June 2019
Several months ago I got a faulty Simpsons Bowling PCB from a friend as a gift. This was just the bare PCB stack in a metal frame and is a
standard Konami GV999 main board with an add-on top board that holds some flash ROMs and associated logic and the trackball controller
IC.
I wanted to test it so I pulled a SCSI CDROM drive from one of my other Konami GV games (Hyper Athlete) and connected it, burned the CD
image from MAME and powered it on. It passed all the tests but came up with a security error and then did nothing else.
I pulled the eeprom (which was curiously in a non-factory socket) and read it then compared it to the dump in MAME and it matched exactly.
This was a bit unusual because MAME eeproms are factory defaulted and so was this one on the PCB, but I assumed it was working at some point
and would have different high score names in the eeprom. At this point I had no other ideas about how to get it going so I put it aside.
Recently I've been doing some repairs to playstation-based Namco System 11 PCBs (repair log coming soon). While repairing them I have been
searching for IC's from other junk boards of the era to fix the System 11 boards. Back in November 2007, I got hold of a bunch of things to
dump. One of them was Tokimeki Memorial Oshiete Your Heart (complete with metal box and CDROM drive), which runs on Konami GV hardware.
While this is playstation-based, it didn't have the correct Sony chips I needed for Namco System 11 but I figured since this game is pretty
useless in it's current form, I would convert it to something more interesting. Maybe even use this board to run the Simpsons Bowling game
if the main board was the problem with it. I used the CD drive and disc from my Hyper Athlete to test the Tokimeki board. To swap games only
requires the CDROM disc and eeprom on the main board to be changed. I didn't have the eeprom dump from Hyper Athlete handy so I downloaded
the one from MAME and programmed it to a 93C46 eeprom. I powered on and it said 'ERROR 25C MBAD', which means the eeprom is bad. I then got
my Hyper Athlete PCB, desoldered the eeprom and read it and was astounded by what I saw. The dump in MAME has had a 16-bit byte swap applied
to it LOL!! No wonder it didn't work! Obviously this has happened sometime during all the (mostly unnecessary) code-playing & shuffling that is happening
nowadays with MAME because I know for sure my eeprom dump from this exact board was correct and that was what was in MAME for years. I re-programmed
the eeprom with a dump read from my working Hyper Athlete PCB and of course it worked just fine, so I know that the Tokimeki PCB
was working. To verify the MAME dump, I loaded it into my eprom programmer software, did a 16-bit byte-swap and it matched my defaulted real
eeprom dump.... SNAP!
Then I started thinking about the error on the Simpsons board. Hmmmm.... it just said there was a 'Security Error'. If you have read my
Konami GV readme in the MAME source (konamigv.cpp)
you will know the eeprom on GV was a mild form of security to stop operators swapping games. The security error suggests the eeprom is bad.
Hmmm I wonder..... I read the Simpsons Bowling eeprom and of course it matched the factory-defaulted dump in MAME... which as I've just
discovered is wrong LOL! Obviously someone has been playing around with something they know nothing about on this board and wrote the EEPROM
from the current MAME dump to the 93C46 EEPROM on the board thinking that would fix the problem LOL! I read the eeprom again, applied a 16-bit
byte swap and then re-programmed that back to the eeprom. Powered on and got this.....
Some progress! Now we are getting closer to discovering the *real* problem. It's telling me the flash ROMs at location 3A, 3B, 7A & 7B are bad. 25C (the eeprom) tests ok now :-)
I powered off and booted up with test and service held in and got this screen.....
After a few minutes the flashing completed then it showed this screen.....
I switched the test switch off and it went directly to test mode. I selected 'Game Mode' and this came up....
It's working! Awesome!
Since it is so rare, here's a large pic of the daughterboard from Simpsons Bowling so you can check it out.
As I have a proper GV metal box (from Tokimeki) I put the whole working board into the box to make it complete :-)
I went back and checked several of the MAME Konami GV eeprom dumps and yep! they are all bad, with an incorrect 16-bit byte swap! Now I'm
wondering what else has been screwed up. Obviously someone who is playing with the dumps doesn't understand the massive effort and expense
that it took to get access to these boards and dump them. In this case no serious and unrecoverable damage has been done. I've informed smf (MAME's psx man)
and hopefully this problem gets fixed quickly.
Regarding repairs to PCBs, just be aware if you are trying to fix a real PCB, you can't always trust the MAME dump of an eeprom, especially
if it was hand-crafted. Just because it works in MAME does not mean it works on real hardware since in MAME the code can be forced to do
whatever the emulation dev wants. A similar issue happened with the dumps of Amiga kickstart ROMs in the past where the dumps were bit-swapped
to keep emulators happy. Obviously this is the wrong way to think... the dump should be kept in its original form and the emulation
changed to suit it. Thankfully the Amiga kickstart ROMs in MAME are good because I pointed out this issue years ago and they were all fixed,
although I don't think WinUAE (the source of the bad dumps & bit-swapping) uses these correct dumps :-/
To those emulation devs playing with dumps, please don't. These dumps took a great deal of time, effort and LOTS of money to acquire and dump
over the last 20-something years. I certainly won't be re-buying things, but it would be a real shame if other people had to find these rare
boards again to re-dump them just because of silly things like unnecessary bit-swapping to keep emulators happy. Think before you bit-swap.
24th May 2019
Following on from the previous repair below, here is the repair log of the other Dodonpachi Dai Ou Jou PCB that came in for repair.
This one was not in bad condition but someone had also removed all of the surface-mounted ROMs (urggh!) and then put them back using little to no skill or technique.
On power-up the screen showed an error 'Trap 13' with some memory registers and other things. Sometimes it would show 'Address Error' but still with the same numbers.
The first thing I did was remove all of the surface-mounted ROMs and check for PCB damage. There was a bit of corrosion around the ROM legs,
and like last time, some surface-mounted pads were lifted by the previous owner and one pad was missing. Fortunately the missing pad is not
used so that didn't affect anything. I cleaned up the board so it was ready for mounting the ROMs back onto the PCB.
The ROMs were read and matched the known good set in MAME so I mounted the ROMs back onto the PCB. I also read both EPROMs and they also matched the ROMs in MAME.
After testing the board again the same error came up on screen but at least now I know the ROMs are good.
The 'trap 13' or 'address error' problem seems to be memory-related. The obvious choice here is to look at the main RAM. I removed the four
256kb surface-mounted RAMs and tested them and found that three tested ok but one tested as bad! I got a new RAM from my stock of ICs and
replaced all four RAMs back onto the PCB.
I powered up and got this.....
Success! It's working!!!
I played a game but noticed immediately there was no sound. I wet my finger and wiped over the back of the board where the power amp IC is and got crackling so the amp chip was ok.
I inspected the board closer and noticed some cowboy has been here before and messed with the 78L05.
This IC in a TO92 package looks like a transistor but is a voltage regulator. It takes 12 volts from the JAMMA connector and outputs clean
5V power for the NEC uPC844C op-amp IC. I measured the voltages and both input and output measured the same. While
this is not surprising in itself if the in/out are shorted, the voltage was very unusual.... it measured -3.5 volts!! Yeah... MINUS!
I removed the voltage regulator and checked it closely and noticed that it was the wrong part!!
It is a 79L05, which is a -5V regulator LOL!!! WOW! What a turkey!
I found a replacement 78L05 voltage regulator in my parts stock and soldered it onto the PCB.
I powered on and got the correct sound.
Luckily the negative voltage didn't damage anything else so the game is now fully working :-D
Here is a very large high quality pic of the whole PCB in its final repaired state. Click the thumbnail then click the arrows at the bottom right corner to expand the pic to full
size. Then you can scroll around and look at the PCB using the mouse by left clicking and holding then moving the mouse.
There are a few important things to take from this repair log.....
A fault on a PCB can sometimes be caused by a relatively simple and common part and easily changed out for a new working part in just a few minutes.
Do not mess with *very* expensive and complicated PCBs if you have no knowledge or skill with surface-mounted repairs. If you want to learn, practice on a junk PCB that is worth nothing so that if something bad happens it doesn't matter.
If you remove a faulty part, be sure you replace it with an identical or functionally equivalent working part!
If you have something like this that needs repair and you don't know how to repair it, don't just hack it up blindly hoping that your
'hammer-approach' will fix it. I can tell you now you can't fix an arcade PCB with a hammer. If you want it fixed, instead of butchering it,
get help from a local friend who has PCB repair experience, or send it to someone who knows how to repair PCBs so that the board will not end up scrap..... especially if it's something nice like a Cave shoot-em-up :-)
