Return Of The Invaders repair log

 PCB Repair Logs, Repair Logs  Comments Off on Return Of The Invaders repair log
Feb 182017

Recently I got hold of this favourite game of mine after a long searching. Apparently it is not so common nowadays, so when I saw a sales thread of this pcb  in faulty conditions from an italian arcade forum I took immediately the opportunity and bought it.

The seller posted a video in which the sprites where wrong and the red and blue components were inverted (this latter problem was caused by wrong wiring of jamma adapter)

When the pcb arrived it was in very good conditions with all rams socketed in factory, therefore I immediately tested them out of circuit which unfortunately  were all good (so no easy repair 🙂 )

It had also an horrible hack on the sound amplifier, more on that later.

Therefore I started the troubleshooting without having any schematics and no similar hardware to look for.

Game had this main fault:

and even if it was not seen in the video of the seller, sometimes I got also the logo and some other images messed up


The sprite problem I was sure was caused by some wrong addressing of the sprite graphics, therefore I shorted some address line of the eproms on the video board to see which ones contained the sprite graphics and found the four responsible.

Addr 7 and 8 of the eproms 10 to 13 were fixed low. I traced back to a 74ls153@IC38  which had both outputs fixed .

Piggybacking it with a good one restored the sprites, therefore I proceeded with replacing it and fixed the sprite problem

The TTL was from Fairchild (not Fujitsu part!) and it was the oldest one on the pcb, dated 1979!

Now the logo problem, it was strange because sometimes it was shown right, sometimes bad.

After some hours of playing, the logo was not shown right anymore, therefore I asked David Haywood (Haze of Mame fame) where it was contained and it pointed me in the eproms A14 and A15 (different bitplanes which were combined to form the final pic). Both eproms had to be addressed in the same time.

Also he told me that the graphic was bank switched, therefore some logic had to select a particular part of the memory area to show the correct gfx.

I probed their address lines and found the upper one A12, was always fixed low. I forced it high during gameplay and the backdround got corrupted but when the title was shown , the logo was there clear.

So the logo was contained on the upped half of the eproms and the logic was not selecting the A12 bit high or low depending on the situation.

I traced back to the only Fujitsu ttl present on the board, a 74LS259@IC72,  8 bit addressable latch.

Before desoldering it, I tried to understand the datasheet of the part, which is not very common to see on arcade boards , and probes 3 select inputs which were oscillating and the enable line was pulsing everytime it changed from title logo to gameplay.

All his outputs were stuck low, and the HP logic comparator confirmed some outputs were at the wrong logic. Only few are used by the hardware, but I expect in later levels I could get other wrong graphics…

I desoldered it and tested out of circuit:

Turned out Fujitsu was again the source of problems!

I had a hard time to find a replacement, because as said the 74259 is very seldom used by arcade hardware.

At least I found one in the following junk pcbs I had: Rock and Rope boot, Equites and Dig Dug.

I took the one from the bootleg (it is always a pity to cannibalize an original board 🙂 ) and problem was fixed!


Since the pcb would go directly in my collection,  the last thing I wanted to fix, was the ugly hack they did by replacing the original amplifier with an LM380 (same used on 1978 space invaders….)


This thing worked but it was really ugly!

After spending one hour to clean the disaster with soldering irom and soldering wick, I installed back in place the 2x missing caps and I looked to find the original MB3731 amp among some of mine junk pcbs.

Besides being used on Taito hardware of the era, it was often used on Konami hardware from late 80s/early 90s.

I took it from a junk Main Events pcb and now it looks much better:


Nova 2001 repair log #1

 PCB Repair Logs, Repair Logs  Comments Off on Nova 2001 repair log #1
Jan 312015

I got this pcb in December 2014 from Smitdogg of Mameworld as not working:

Foto 10-01-15 20 05 02

It was in good condition but it had already some TTLs replaced (socketed)

After firing it up I got a static coloured screen.

The Z80 was socketed so I first tried to replace it and to my surprise the game started apparently without any major faults with correct colours.

It was too good to be true and infact when the demo play started I got a very peculiar fault:


The background colours changed alternating between correct and not only during horizontal scrolling. During vertical scrolling the colour of the background was fixed in the last state (being good or bad).

I started the test of the pcb and everything looked good except the color check.

The “Back”  had some strange colours which didn’t look correct



I decided to start Mame and see the same screen on the emulated game and I got this:


it seemed that there were some fixed bits on the background colours but that didn’t explain the alternating colours.

I looked for Nova 2001 schematics which luckily were available.

This game has plenty of 4116 drams which are prone to fault normally but they weren’t related to the fault I had.

I probed also the 3x  6116 SMD rams responsible for the background colors and the related 74LS273 buffers and they were all good.

At this point I had no clue where to probe, the schematics were really bad scanned and difficult to read.

I asked an advice to Charles MacDonald which pointed me to look on the AY8910 sound chips whose I/O are responsible for the  horizontal and vertical scrolling.


The first 3 bits BH0, BH1 and BH2 of AY8910@1,2A are negated and sent to a series of 74LS151 @12U, @11U, @10U and @9U.

All the signals going to these 74LS151 were good.


I probed the outputs of all the 74LS151 and noticed that the TTL @12U was oscillating in sync with the alternating  background colors.

After checking the inputs of this chip, I noticed 3 stuck signals which were back traced to the 74LS164@12T whose inputs CLK and CLR were correct.

Changing this TTL restored the correct colour of the background during scrolling.