Repair Logs – Page 100 – JAMMArcade.net

Rainbow Islands repair log #6

Jan 212015

All good things come in threes!

Yes, this is the third Rainbow Islands PCB (the fourth if we include the Extra version) I repaired.The PCB was in good state (dust apart):

But all I get was a solid blank screen upon powered it on, no activities on DATA/ADDRESS bus of 68000 CPU, RAM and ROM.The program ROMs were dumped as good so I went to the two 6264 WORK RAMs @IC25 and @IC26 and, mindful of what had already happened with the other Rainbow Islands PCBs, I desoldered them and tested out-of-circuit.Both were bad!

With new WORK RAMs fitted the board sometimes booted stuck on “OBJECT RAM ERROR” other times with jailbars all over the sprites:

Jailbairs are a clear symptom of wrong written data.Sprite generation circuit on this PCB is made of a PGA custom ‘PC0900J’ , four 2018 SRAMs , some latches and buffers.Piggybacking the RAMs had no effect but when I made it on a 74LS244 @IC15 the jailbars faded and then disappeared completely when I piggybacked the other 74LS244 @IC17.

Desoldered them and tested out-of-circuit confirmed they were both bad:

Graphics were perfect at this point.Last issue to troubleshoot was the absolute lack of sound.When I was gonna to analyze the sound circuit I noticed @IC44 the 6264 RAM of the Z80 audio CPU was already socketed so I decided to test it resulting in a bad chip.Replaced it restored the sound.Board 100% fixed!

NEC PC-9821AP2/U8W repair log

Jan 182015

There are not only arcade boards to repair in the life…but also computers (and japanese ones are really cool)!

I got from Japan this ‘junk’ NEC PC-9821AP2/U8W in the mail some days ago (after a wait of two months), seller said only it was not working without going into details.

Powered it on and I got a solid black screen so was time to disassemble it.Motherboard was clean except for this:

Surpringsly all electrolytic capacitors were in good state for a machine with more than 20 years old but there was sign of corrosion near five 22uF 16V tantalum capacitors ~~which leaked spreading their dielectric on pcb.~~This was most likely caused by the glue used to hold them in place that turned corrosive with age.These capacitors were mounted as decoupling on five 74F245 which are used for DATA communication between DATA buses.In this hardware they were interposed between four NEC 42S4800 DRAM chips and connector of the CPU riser card.So, with these premises, I was quite sure that fault was located in this crucial part of the motherboard.After removing the ~~leaking~~ capacitors and cleaning the circuit, I started to probe continuity between each pin of the five 74F245 and motherboard and , following a scheme that I had prefigured, I found that PIN2 of one of these was not connected to the CPU card connector like all other so one bit was missing and the system was halted.Jumpered the two points and finally system successfully booted:

But immediately something sounded strange to me.According to its specs the PC-9821AP2/U8W model should have 5.6MB of default RAM (first 640KB are conventional memory plus 5MB of extended RAM) while, as you can see from picture above, mine had only 3.6MB so I missed 2MB of them somewhere.Also using an addon RAM card didn’t change the amount of memory.So, something else had to be wrong.Luckily I had another same PC with a good motherboard and tracing it I found that the five 74F245 had PIN1 (direction PIN which select the direction of the DATA transfer) in common while in the faulty motherboard this was only for two of the TTLs.So, due this, DATA could not be transferred from all available RAM to CPU resulting in only 3MB in total.

So I tied PIN1 all toghether with some AWG30 wire:

and the missing 2MB came back as well all extra memory of the addon card:

Another job done.

G.I. Joe repair log #2

PCB Repair Logs, Repair Logs Comments Off

Jan 142015

Yes, yet another G.I. Joe PCB on the bench (this is the third I repair).Let’s start with the log.

PCB was in a clean state:

But, as often happens with this Konami hardware, board had the watchdog active sign that there was a problem in main code execution or in the CPU itself.As usually I started my troubleshooting analyzing the main 68000 CPU and I found that, though the RESET signal was correctly generated by the custom ‘051550’ and reached the CPU, PIN17 (/HALT) and PIN18 (/RESET) of 68000 were stuck LOW so it was not never inizialized (all control lines were obviously silent ,too).So, judging from this,I started to suspect that 68000 was really faulty.I decided to do an “extreme” piggybacking with a good know CPU and sometimes I got it booting showing a bad color RAM:

This was enough to desolder the main 68000 CPU (a 16MHz one manifactured by ST) fitting a proper 64PIN socket:

Trying it in another board I obtained a solid black screen sign that it was realy bad (it’s the first time I came across a faulty 68000 CPU, never happened before).With a new 16MHz 68000 fitted in the socket board succesfully booted always showing the color RAM error above but I quickly fixed this by replacing both 2018 SRAM (also the one @14C was bad)

As I expected in this kind of Konami hardware which use ‘054986A’ hybrid custom sound module there was absolute lack of sound ,only some popping noises sign that the LA4707 amplifier was still alive at least.So was time to replace all the SMT electrolytic capacitors with new tantalum ones as they were leaking and having ESR value out of specs:

and install proper 1.778mm pitch sockets for the module to finish the job:

Son Son repair log

PCB Repair Logs, Repair Logs 1 Response »

Jan 122015

Welcome to my first repair log!

I bought this (very) expensive pcb from ebay in October 2014.

I played the game a couple of times and then after about one week, all the sprites disappeared while I was playing. Only background and text were present:

I couldn’t believe, at first I began to press everywhere hoping it was a loose or oxydized connector but soon I realized something got faulty.

This game hasn’t any schematics available so I knew from the beginning it would have been very tough to fix it.

I connected my trusty logic probe and began to short some pins on the pcb to see what changed on the screen so that I could focus on the right part of the circuit.

After about 30 mins turned on, the sprites started to reapper but missing some lines:

I then remembered that 1942 hardware is very similar to Son Son and the schematics are available.

I was right, the way that TTL chips are connected on 1942 is equal on Son Son. Only the positions of the ICs are different but the logic is really similar.

I probed the 2148 rams @F2 and @F4 on Son Son pcb which on 1942 schematics are described as OBJ rams and I found a stuck /WR signal @F4 on Son Son (J4 on 1942):

I followed back the signal always taking a look on 1942 schematics, and I found finally the source of the problems on the missing signal on pin 11 of an 74LS00 @F11 (on 1942 I circled the equivalent IC @K11).

On 1942 schematics available on internet, someone circled the 74LS00@N8 writing “no plane”, probably he had the same fault. 😉

Replacing the 74LS00@F11 gave me back all the sprites

Lightning Fighters repair log #2

PCB Repair Logs, Repair Logs Comments Off

Jan 062015

Found this Konami Lightning Fighters PCB buried somewhere in the pile so today I decided to take a look at.

I know quite well this kind of hardware since I repaired two of them in the past and I can say it’s not really complicated, the use of graphical ASICs simplify a lot the PCB layout and troubleshooting.

Once powered on I got a solid black screen.Probing the main 68000 CPU revealed that watchdog was active and board was resetting endless.Dumped the two program ROMs gave me bad dump of the one @E15.Once reprogrammed a 27C010 EPROM, board succesfully booted but all colors were clearly wrong:

Schematics of this game were available so looking at them revealed that color RAMs were two 2018 @D18 and D19.Probing these two static RAMs I found some addressing lines not pulsing fine.All of these were tied to the four outputs of a 74LS157 @F19 (whose inputs were connected to address lines of 68000 CPU and ASIC marked ‘053251’ so this was how the two colors RAM were being addressed) .

Piggybacking a good 74LS157 restored all colors.Using my HP10529A logic comparator was a further proof of my suspects that were confirmed definitively testing this multiplexer out-of-circuit where it failed miserably:

Fitted a good 74LS157 brought back the PCB to 100%.

Older Entries Newer Entries