Donkey Kong 2-Board Missing Colors Issue (FIXED!!! 9/3/08)

[DonkeyKong]

UPDATE:  FIXED!!! on 9/3/08!  Read the entire thread for all of the little debug bits... and here is the FIX post:

http://forum.arcadecontrols.com/index.php?topic=83374.msg879867#msg879867

********************************


Hey guys and gals!

I'm currently stumped on a problem I have with my 2-board DK set.  When I first got the board, all of the colors worked fine... but only after playing it for awhile did the colors start to "disappear"!  Certain colors have disappeared, but only on certain screens.

I have a 4-board set and replaced all of the ROMs and PROMs from the 2-board with the ones from the 4-board, with no change.

I bought one of the Brasington High Score Save Kits (D2K pre-installed!) and the self tests report that all of my Video and CPU ram and rom is good!

Please take a look at the following screenshots and let me know if you have any ideas what the problem could be.  I'd really appreciate it!!

In particular, colors are missing from Donkey Kong, Jumpman, The Lady and Fire on the barrels screen.  Also some weirdness on the colors above the lady.

I have access to oscilloscopes and DMM's and what not, if you want me to look at something specific.  Please don't hesitate to get technical   I'm dangerous like that...













THANKS IN ADVANCE!!!!!

[Paul Olson]

I don't see anything that looks exactly like that, but check here if you haven't already.  http://www.brasington.org/arcade/tech/dk/

[DonkeyKong]

Yes, I should have said I checked there already.  That's a good resource if one of those chips is bad.  However, my HS save kit says everything is good.  Whaaaaaa!!! 

[DonkeyKong]

I want to work on this thing but I'm not really sure where to start.  If all of the RAM and ROMs are reportedly working, I would think it's all good, but maybe there's a problem between the video output and video RAM??

DK

[djruden]

I hate to sound obvious and I know these options probably aren't available, but firstly I would try to find another DK board and swap out to make sure the problem is on the board itself (ie a chip is bad). 

If it's not the board then it might be the cable running from the boards to the video doesn't have a good connection thus missing some of the video output. 

Monitors themselves are a weak point for me tho so I'm not how to approach a problem there.

[DonkeyKong]

It's not the monitor, because it works fine with my Donkey Kong 3 board set.

If I was going to buy another 2-board set, I would just send in mine for a working one from mikesarcade.com at this point.  I've spent so much already though that I cringe at spending another $85.  I really want this game working though!  It's no fun practicing DK when it looks like that ^^^

[SirPeale]

First, make sure all your socketed chips are in the sockets correctly with no pins accidentally left out or broken off.

I'm not sure which color you're missing there, actually looks like a combo of colors.  So I'm leaning toward one of the PROMs being bad.  Which one, however, is the key. 

[SirPeale]

I would guess 2F, but that's just a guess.

[cornishHen]

I'm in a very similiar boat.   I'm missing some color(s) as well.  I have a d2k card, but the game had this problem before the d2k board.  Also, all my rams from the d2k self test show as OK.  The monitor i'm using was working on a dk3 board beforehand.  I bought this board from a member on klov.  I received it and had some problems with the whole screen being scrambled. I sent it back to him, but he stated it worked fine on his bench test so he sent it back to me.   I then found I had a bad connection on my switching power mod.  Now I stand with the situation below and hoping to fix it myself as shipping this thing around is expensive.   All my voltages appear to now be correct ( checked them 10 times atleast).  not sure what to try next..

[DonkeyKong]

So I tried what Peale suggested... I removed the PROM from 2F and scraped the pins clean.  These smaller PROMs are the only components on the board that had very tarnished pins (blackened).  I had previously cleaned them, but scraping them with a small flat head screwdriver worked really well.  When I put the chip back in I was being distracted at the time and put it in BACKWARDS!  When I turned on the machine the chip smoked and the game was in a funky loop displaying scrambled characters and looping a sound, much like my 4-board problem.  I'm not sure, but I think the 5V supply at that point could have been pulled down.

Luckily I had the 4-board set to the rescue and stole it's 2F equivalent PROM.   I cleaned the pins the same way and socketed the chip.  I was back up and running, but the colors we SLIGHTLY different... there were a couple little specs of color in Jumpman now and maybe more color in the Fire.  When I remove 2F and boot the machine, the colors get really weird like shown on the Brasington site.

I think I may try to get some new PROMs and program them with my EPROM/PROM burner... or just order them from someone.  Anyone know where I can go for these PROMs?

[DonkeyKong]

Ok, a set of 3 programmed DK PROMs can be purchased here:
http://www.mikesarcade.com/cgi-bin/store.pl?sku=DKPROMS

This is a steal considering out of all the cross references, the NATIONAL 74S287 (which is equivalent to the TI 24S10), is $12.95 EACH here (and still require programming):
http://www.jameco.com/webapp/wcs/stores/servlet/ProductDisplay?jameco_page=42&langId=-1&productId=48696&catalogId=10001&freeText=74S287&storeId=10001&search_type=all&ddkey=http:StoreCatalogDrillDownView

DK ROMSs here:
http://www.mikesarcade.com/cgi-bin/store.pl?sku=DKROMS

I swore I used my programmer to verify the PROMs and they were fine... I'll have to check that again.

[cornishHen]

Not trying to hijack this thread.. I just thought my problem was quite similar. 

DonkeyKong sounds like your on the right track.. I wish I had some spare pcb parts to work with.  I do have a EPROM programmer and may have found the cause of my missing colors, but I wanted to get some opinions..

I decided to run romident on all the EPROMs from the video board ( from 2 board pcb) today.   I ran into one -   7F  that showed as "Not Found!" in  winromident.

dk7dt.BIN: (1/1) [672E4714]
   - DK.7D : Donkey Kong (Nintendo)
   - 4N.CLK : Donkey Kong (Nintendo, jpn)
   - l_4n_b.bin : dkong:Donkey Kong (US set 1):Nintendo of America
   - l_4n_b.bin : dkongo:Donkey Kong (US set 2):Nintendo
   - l_4n_b.bin : dkongjp:Donkey Kong (Japan set 1):Nintendo
   - l_4n_b.bin : dkongjo:Donkey Kong (Japan set 2):Nintendo
   - l_4n_b.bin : dkongjo1:Donkey Kong (Japan set 3) (bad dump?):Nintendo
============================
dk3Pt.BIN: (1/1) [15E9C5E9]
   - DK.3P : Donkey Kong (Nintendo)
   - v_3pt.bin : dkong:Donkey Kong (US set 1):Nintendo of America
   - v_3pt.bin : dkongo:Donkey Kong (US set 2):Nintendo
============================
dk7ft.BIN:[C4E728E0] Not Found!



I thought this was odd and wondered if maybe winromident just didn't have the correct database entry for that particular rom.  As I don't have any clean roms handy for programming, I decided to put all the roms except for 7F back into the video pcb and crank it up.  The game looks and behaves( plays but missing colors) with the missing rom..Is it safe to say there is most likely a problem with 7F?   

   

[DonkeyKong]

Hey CornishHen,  no problem adding your particular issue in this thread... the more the merrier!  If one of us finds a fix, hopefully that will help the other   It does look like your issue is very similar to mine.  What I noticed with my problem, is that over time the colors disappeared... like BIT FADE.  I suppose it could be a particular bit on a data line that is toast.

I've done some experimenting in MameDK (Tiny Mame built for DK), and found that changing the various bytes of 7F seemed more likely to cause the problems we are seeing.  I also tried doing similar tests to all 3 PROMS and they have much different effects.

Sounds like you used Romident on the read back values of all of your EPROMS and ROMS... and if Romident is saying 7F is completely missing.. it sounds like you read back all FF's for that chip.  This would be the same thing the system sees if the chip is missing from it's socket.

If your machine looks exactly like this with 7F missing, it most likely is 7F.  You could try erasing and reprogramming 7F first, but it's unlikely to work if that chip is bad.


I'm going to try removing 7F on mine to see what it looks like.

[DonkeyKong]

Well removing 7F seems to resemble that image with my board, but I still have the same missing colors.



I tried cleaning a little more, and reseating the 50pin cable connectors about 10 times each.

My problem is weird, the missing colors are not stable, they are always fluttering on and off a little bit.  Maybe I'll make a video of this at some point.  CornishHen, do your colors flutter at all?

Also, I guess the Brasington High Score board does not verify VIDEO ROM... only PROGRAM ROM.

[DonkeyKong]

SWEETNESS!!!!

OK, finally some progress has been made on TWO fronts!

I found a program called WinHex that would allow me to do bitwise operations on the ROM bytes, which is helpful in simulating floating/shorted pins of the various EPROMS/PROMS.  I figured out the right combination of bit value and chips, and when I take the ROMs at 7C and 7D and set them to ALL ZEROs... I resulted in the following image in MAME.  Keep in mind that when simply removing the chips one at a time (See All Pics on the Brasington Site), the equivalent would be a chip programmed with all FF's.


Compare that to my machine...


Now my machine MOSTLY has ZEROs for these locations... it somewhat flickers the right colors but for the most part they are zero.  Now I'm pretty sure I verified these EPROMs before with my programmer and they were fine.  This leaves errors in addressing these chips and transferring the data to the monitor, which would involve the video RAM.  Personally I would think the RAM is highly suspect, especially since the colors were good and then faded away.  However, the BRASINGTON HS KIT says VIDEO RAM is OK.  Well, somethin' in there is busted.  Now I know where to zero in on!  It's just a matter of time now before it's completely fixed.

So my second success was finally figuring out how to repeatedly perform a successive jump action, and BETTER control the barrels.  This jumping was really holding me back when I would get 2 or 3 barrels fairly closely spaced, and tried to jump quickly with nothing happening.  Now I know the secret   I figured these things out during the 35 simulations I ran today.

With this new information at my fingertips, I set out to play a game on my machine and finally log a respectable high score in the Brasington HS Kit.  My previous high score was 102,900 - and I shattered it IMO tonight with 145,200!  This was getting past the 4th elevator stage, and 5th conveyor belt stage.  The jumping and barrel control was really helping!

[richms]

Hows the power to those sockets? I have seen noisy trash out of roms in other things when they had lost either vcc or ground, and were reliant on the protection diodes and io pins to run the chip. Worth a look anyway.

[DonkeyKong]

Hows the power to those sockets? I have seen noisy trash out of roms in other things when they had lost either vcc or ground, and were reliant on the protection diodes and io pins to run the chip. Worth a look anyway.

Hi Richms,

Thanks for the reply.  Better yet I measured the voltage with the EPROM's in-socket fully loading down the 5V bus voltage.  The highest and lowest voltage my DMM measured was 4.92 to 4.97 volts.  I didn't have a scope handy, but the peak hold on meter works pretty good for now.  The resistance from the bulk capacitor just above that bank of EPROMS was good and low to all of the video EPROMs.  I'd say the power is good.

Another interesting test was REMOVING the 7C and 7D EPROMS which should cause the system to read all FF's instead of whatever the EPROM contains.  Comparing this with a MAME simulation of the same test, we see that my system mostly still reads zeros and only a few high levels at various bits.





Something else I am looking into... are the two RAM modules on the Video board that have the heatsinks on them.  These are MB7074's located at 2E and 2H.  These get video data directly clocked into them from the video EPROMs, through various logic, but only clock in data from those EPROMs or from themselves off a latch.  I'm really having a hard time believing it's anything but RAM or ROM... and since I know it's not ROM... it must be RAM.  The logic chips that shift in data wouldn't be flakey, they would fail a good high or low on some input/output and would put out a very deterministic pattern.  Those heatsinked RAM modules are highly suspect because they ARE HEATSINKED!!  Maybe I'll try some cooling spray on them and see if that makes the random bit twiddling of disappearing color stiffen up for a little while.

[cornishHen]

Thanks for all the info/research Donkey Kong. 

 I received some blank 2716 eproms in the mail today.  I keep getting an error when attempting to program the blank chip to 7F..

I have the programmer pictured here:
http://cgi.ebay.com/Newest-Version-PCB6-0-Willem-EPROM-Programmer-BIOS-PIC_W0QQitemZ250287689898QQihZ015QQcategoryZ4661QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
.  The programmer is a dual power deal USB/AC adaptor.  I've tried both methods of power USB and 12V with no luck.  All the jumpers/dips appear in the correct positions.   I keep getting "Error at 0X000000 Chip 0xFF".  Anyone have tips of what I may try?

[DonkeyKong]

Thanks for all the info/research Donkey Kong. 

 I received some blank 2716 eproms in the mail today.  I keep getting an error when attempting to program the blank chip to 7F..

I have the programmer pictured here:
http://cgi.ebay.com/Newest-Version-PCB6-0-Willem-EPROM-Programmer-BIOS-PIC_W0QQitemZ250287689898QQihZ015QQcategoryZ4661QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
.  The programmer is a dual power deal USB/AC adaptor.  I've tried both methods of power USB and 12V with no luck.  All the jumpers/dips appear in the correct positions.   I keep getting "Error at 0X000000 Chip 0xFF".  Anyone have tips of what I may try?

I don't know if you have these options, but you should just double check that you are setting the programming range from 0x0000 to 0x07FF, set the programming voltage to 25V, and use "l_4s_b.bin" from the dkong.zip file as your program memory for the 7F video eprom.

Try to do a blank check on the chip to make sure it's completely erased to all 0xFF's.  If it's not blank you'll have to get a UV burner and erase it first.

I'm just guessing but I think you are going to have to definitely plug in the 12V power supply.  5V from the USB port might have enough current to boost 5V to 25V, but I don't know if it actually does that.  I know your board supports a 25V 2716 eprom though.

That's about all the help I can offer, since my programming system is ancient and running still on a Pentium 75 from DOS.

[cornishHen]

Found my problem with the programmer.  A jumper setting for the voltage.  I jumped it to the correct 25v setting and the eprom programmed beautifully.  I went ahead and ran it through romident and it recognized as 7F.. But, after placing it in the game there was still no change.    Back to the drawing board for me....

[DonkeyKong]

Get on the bus Gus!  I have my oscilloscope powered up right now and I'm looking at the path between 7C/7D and Video RAM.  I'll update later with details.

I had spent some time thinking about your color situation.  Your colors seem very stable.  I would think you have a cracked trace or shorted node somewhere... I tried playing with the color mapping PROMs and couldn't generate that exact look.  Maybe when I get mine working I can be more helpful!

[Paul Olson]

cornishHen - Do you think that programmer will work for all of the chips on a Nintendo board? I have 4 boards that need some work, and at that price, I could save some money. I guess I need to figure out how to erase them too.

Thanks,
Paul

[cornishHen]

cornishHen - Do you think that programmer will work for all of the chips on a Nintendo board? I have 4 boards that need some work, and at that price, I could save some money. I guess I need to figure out how to erase them too.

Thanks,
Paul

It should work for the 2716/2732 eproms... I think DK also has 2532 eproms?   I'm not sure if/how it handles those.

[cornishHen]

Get on the bus Gus!  I have my oscilloscope powered up right now and I'm looking at the path between 7C/7D and Video RAM.  I'll update later with details.

I had spent some time thinking about your color situation.  Your colors seem very stable.  I would think you have a cracked trace or shorted node somewhere... I tried playing with the color mapping PROMs and couldn't generate that exact look.  Maybe when I get mine working I can be more helpful!

I am starting to wonder if my power solution is wrong..  I am using a power wiring harness that was originally on a DK3.  It uses 1 large edge connector to connect to the boardset.  I converted the power end to work on a switcher based off a DK edge connector pinout. 

I keep seeing dk harnesses on project pages that have the 3 connectors hooked to the  dk board.    I see that arcade shop sells them for a switcher: http://www.arcadeshop.com/pics/kong-ps-kit.jpg .

Is that the preferred method for supplying power as opposed to the edge connector on DK boards?

[DonkeyKong]

I am starting to wonder if my power solution is wrong..  I am using a power wiring harness that was originally on a DK3.  It uses 1 large edge connector to connect to the boardset.  I converted the power end to work on a switcher based off a DK edge connector pinout. 

I keep seeing dk harnesses on project pages that have the 3 connectors hooked to the  dk board.    I see that arcade shop sells them for a switcher: http://www.arcadeshop.com/pics/kong-ps-kit.jpg .

Is that the preferred method for supplying power as opposed to the edge connector on DK boards?

I think most people just use 2 power connections for the 2 board sets, and 3 connections for the 4 board sets because that's the way the older machines were wired.  I don't think until DK3 did they start truly using the edge connector.  I could be wrong about this though so don't take my word for it.

Based on your screen, it looks like your power connections are fine.  I think you may just want to double check the monitor connections, even though I highly doubt one of them is bad - or you would not be producing RED, BLUE or GREEN color at all.

[DonkeyKong]

I left my machine on all night long and went to play a game just now...

I noticed that all of my random bits of color were completely gone, and all that was left was exactly like this MAME image:


Things are starting to solidify.  The warmer it gets (and over time) the more color generated from EPROMS 7C and 7D that is fading away.  The color fade seems permanent though... so if I cool things down it doesn't come back.  I know my EPROMS are good though.

Since I've been checking out the signals on the oscilloscope, I've noticed data high and low being clocked out of all 4 video EPROMS, data is getting through the LS299 universal shift registers, through the first LS157, through the second LS157, and through the 10124 TTL to ECL converter.

This is where I stopped because I've never messed with ECL levels before, so after some googling and wikipedia searching, I believe I now know what I should interpret as logic high and low to the input of RAM.

Now I can verify the input and output of RAM, and it's a good thing data has a separate path in and out of this ECL RAM, so I don't have to gate my signals to interpret what is what (like a common I/O data bus).  I'm hoping that's the problem because Mike's Arcade has that RAM pretty cheap.

I'll update tomorrow after I take a look!

[DonkeyKong]

Now I can verify the input and output of RAM, and it's a good thing data has a separate path in and out of this ECL RAM, so I don't have to gate my signals to interpret what is what (like a common I/O data bus).  I'm hoping that's the problem because Mike's Arcade has that RAM pretty cheap.

OK It's definitely the RAM at 2H.  All of the ECL levels coming in on pins 6, 7, and 19 look good, but the outputs are a different story!  Pins 4 and 23 look good, but pin 2 is bad... well sort of.  The ECL levels reach the value considered ZERO logic level, but they don't reach the HIGH level!  That's IT!!  I'm sure before I left it on all night last night, some bits were reaching a HIGH level, but not anymore.

To make sure it's not something dragging on that pin, I've ohmed out and diode checked in every direction to every bus, that pin verses the other RAM outputs.  They all look the same!  So since RAM is the actually DRIVER and the 10125 ECL to TTL chip is just a high impedance input, it's gotta be the RAM.

I'll be placing an order today for some RAM at Mike's and let you guys know how this story turns out after the repair.

BTW:  Tech note, bit D0 on the 2H RAM is responsible for storing the data streaming from video EPROM 7C and 7D.  When it's output is consistently low, those colors go BLACK (or OFF) on the characters.

[DonkeyKong]

So I got my RAM from Mike's Arcade as quickly as possible and installed it tonight.  Thanks Mike!!!

Here's a picture of the install:


I used a 1.125" length piece of 1/2" aluminum U channel to create a new and improved heatsink.  I attached it to the RAM with acrylic adhesive thermal conductive tape.  You can get it at McMaster-Carr part number 6838A11 (1" wide x 12 ft.).  Unfortunately it's $54 per roll!  Maybe if someone needs some I can piece it out

After buttoning it all back up I flipped on the power switch, and low and behold...


IT WORKED!!!!   I don't think I've ever been more excited to fix something in my life.  I was confident that it would work, but hey... IT REALLY FREAKING WORKS!!!!

Immediately after freaking out I played a game and cracked 103K something...  it looks so good now that I really want to play instead of writing this!

So before i go... here's a little picture of a useful thing I did to help me get access to the board while I was probing around with the oscilloscope.  I put two small (I wanted larger ones) picture hanging type hooks into the monitor support, to hang the board off of.  I could leave all of the power plugged in, and now everything was withing reach.


Also, if anyone is interested, here are two 10 megapixel images of the Donkey Kong 2 board set, CPU and VIDEO side.

Donkey Kong 2-Board VIDEO

Donkey Kong 2-Board CPU

If anyone has any questions about all of this, I'll try to help!  I also have a working 4-board set on the way to me, so I can finally try to fix my non-working 4-board set.

[pcb]

Nice Job!  Glad the rams helped!

 - Mike -

[cornishHen]

Congrats DK... I'm just now getting back to my issue.  I've been involved in installing a new hardwood floor in the house that has taken more weekends than I expected...  I decided to measure the B+ on my sanyo monitor today to just cover bases.  It turned out to read 91V on average.   This is well below the 108 needed and might at the least explain the minor wavyness of the screen.  I tried to adjust at the B+ pot but that pot is complete toast.. spins and no change in voltage.. also a bit mangled.  I tried looking at the schematic to determine a suitable replacement, but I can't find the darn thing on the schematic.   

I'm creating a new thread on the monitor repair section asking for help identifying the B+ pot..

[cornishHen]

Back at it again..  Still same problem but have some findings.   I decided to try the blank rom test in mame to try and mimic my problem.  Using a blank 7E and 7F together produces the problem in mame exactly.  I take both of those roms out of dk boardset and no change.  So, it seems the 7E /7F are not being used/recognized.  I reburnt both eproms, verified in romident, and but back in boardset..same problem.  I've poked around on both eproms with the multmeter and they are getting power.  I just got a logic probe yesterday( really don't know how to use it) and poked around..there appears to be pulsing hi lows on many of the eprom pins of 7E/7F.  Besides resoldering new sockets i'm not sure where to go with this.  Anyone have any ideas on this one?  I've added images belowing showing the boardset problem and the mimic'ed mame version.

Thanks!