weird Galaxian sprite dropping fault! help!!!!

andrew96

New member
Joined
Feb 12, 2011
Messages
107
Reaction score
2
Location
Kent, United Kingdom
Have a very weird sprite fault when they drop down, has anyone seen this before or know what causes it?

here is the events in attract mode when first switched on


1st one drops down on right hand side, all ok

then on left side 2 red ones and top row one drop down, end red one disapears right away and the other red one totally disapears while passing 2nd row from top while dropping, but reapears, when they rejoin the pack end red one is there again

right side 2 red ones and top row one drop, one red one on the end of the 3 dropping disapears the moment it moves to drop!, then the 2 left that can be seen BOTH disapear while passing 3rd row from the top red one reapears, but one from top row disapears not to be seen again untill it reapears back on top row after the dive

next dive, ok

next dive again, ok

totally stumpted! I still not sure if problem is with attack ram or object ram, both have been replaced and object ram throughly tested with the fluke 9010a, attack ram replaced along with all buffers, scope shows no stuck lines or anything silly like that.

anyone with any ideas?

it seems when 2 drop its fine, but the problems appear to be when 3 drop in a bunch

this is taken from the attract mode, but play mode is same

all roms have been changed for good ones also, just to illiminate any problems with code

HELP!!




here is a link to the video, i think it has to be downloaded first, but its 990k, i couldent find a way of embedding it, just right click and save as when over the link below


http://www.andrewfinn.co.uk/galaxian.wmv

any help will be good!!
thanks
Andrew
 
Last edited:
From the video, it looks like a fault that I've seen (and repaired) before. Here's the entry from my repair log:


Symptom: Sprites disappear behind invisible horizontal bands (three of them) – one in the middle, one about a third of the way up, one about a third of the way down (note: this board is running Moon Cresta, not Galaxian!).

Cure: Replaced the 74LS32 at 3S (pin 11 was stuck Low) – also worth replacing the LS02 at 3R if this doesn't work
 
From the video, it looks like a fault that I've seen (and repaired) before. Here's the entry from my repair log:


Symptom: Sprites disappear behind invisible horizontal bands (three of them) – one in the middle, one about a third of the way up, one about a third of the way down (note: this board is running Moon Cresta, not Galaxian!).

Cure: Replaced the 74LS32 at 3S (pin 11 was stuck Low) – also worth replacing the LS02 at 3R if this doesn't work

Many thanks Solder

I have indeed replaced both those ic's 3S, 3R is all working, nice clean transisions on both. The odd thing is it is only 1 band, and then it seems only when there are 3 dropping down things disapear and act weird.

I am guessing here now as running out of ideas! but am thinking that a enable signal somewhere is not long enough to allow 3 character positions to be written hence the loss of characters, where 1 character writes fine. i might be totally wrong here but as the dropping characters are stored in the 27LS00 and all are affected, it cant be them! also they have been replaced with new good ones, have changed pratcially everything in the path, but on scope nothing seems wrong!

partial list of changed bits

all 27LS00 ram
3T, 2T ram address buffers
5T, 4T address line generators

3S, 3R, 2P, 2N, 3P, 2S, 2R, 3M, 1U, 1E, 6N logic gates

6P, 6R

2H, 2J, 2K, 2L, 3L, 3K, 3J, 4L character rom interface

6M, buffer to colour prom

I think that covers most of the attack ram circuit!

also 8M and 8N as they generate some of the control signals

its gotta be something i am overlooking!

Am i right in thinking the 27LS00 are OVERLAID on top of the main screen comming from the object RAM character roms? If i remove a character rom when they dive there are then squares, but oddly they to disappear insted of dropping down the screen, so my question is, is the 27LS00 data overlaid on top of the romdata, or is there a hole being cut out of the background allowing the character dropping to show? i am not to clear how that bit works??

many thanks
Andrew
 
I think that your only option is to keep poking around - it's a pain of a fault to diagnose though, might be worth trying to halt the CPU as it happens, then maybe even try piggy-backing known good chips onto those that may be causing the fault.

I wish you luck! Let us know how it goes. :)
 
I think that your only option is to keep poking around - it's a pain of a fault to diagnose though, might be worth trying to halt the CPU as it happens, then maybe even try piggy-backing known good chips onto those that may be causing the fault.

I wish you luck! Let us know how it goes. :)

yes its a real beast!!

I have found a way to freeze it with the fault showing, with a fluke 9010 conected run UUT then just as things disappear hit the bus test, this then freezes it at exactly the right point! (obviously getting the fluke to ignore the watchdog) when character was disapeared i did find if i shorted a couple of lines together on the 27ls00 ram address latches ( 4T ) it moved the character along the line but was then complete again while the lines were conected together, but only half complete figures, missing ones stayed missing! it just has to be something with the 27ls00 drive circuits, but i am stuck which way to go now.

if i find it i will let you know, spent hours and hours on it so far. i wish there was a way of writing to the 27ls00 but cant see how with the fluke so far

Cheers
Andrew
 
Good to hear that you have a Fluke, that will help. :)

its not helped me so far!!! i thought it would be soooooo usefull! but so far all its done is check the ram, which is all ok! was hoping i could find the addresses to turn things on like sounds on the galaxian boards, but no luck so far
 
Well after nearly a year I have finally got back to this board, and wooohooo!!!! I fixed it!!!

Its a bootleg board and does not have the buffer at 1U (74ls174). so after alot of poking and prodding the board I found that the normal 'open collector' sprite memory was fitted (with 1k pullup resistors fitted), but found the output did not reach a digital 'low' sometimes and that was when the sprites went missing! so decided to get the 'tristate' output memory chips (SN74S201 Texas Instruments) that way the output is pushed and pulled between 0v and 5v and of course removed the pullup resistors.

RESULT that cured the fault!!!

So thats the answer! genuine boards have open collector' sprite memory with pullup resistors, and bootleg ones have 'tristate' memory and no pullups and no 74ls174 at 1U

On closer inspection it does appear the resistors are not orignal so do suspect the sprite ram has been replaced in the past, pull up resistors added as this was the norm, and the problem I see had gone un-noticed!

hope that hopes someone at somepoint!!

cheers
Andrew
 
Last edited:
Congrats and thank you for posting the fix. I am sure this will help others in the future.


Well after nearly a year I have finally got back to this board, and wooohooo!!!! I fixed it!!!

Its a bootleg board and does not have the buffer at 1U (74ls174). so after alot of poking and prodding the board I found that the normal 'open collector' sprite memory was fitted (with 1k pullup resistors fitted), but found the output did not reach a digital 'low' sometimes and that was when the sprites went missing! so decided to get the 'tristate' output memory chips (SN74S201 Texas Instruments) that way the output is pushed and pulled between 0v and 5v and of course removed the pullup resistors.

RESULT that cured the fault!!!

So thats the answer! genuine boards have open collector' sprite memory with pullup resistors, and bootleg ones have 'tristate' memory and no pullups and no 74ls174 at 1U

On closer inspection it does appear the resistors are not orignal so do suspect the sprite ram has been replaced in the past, pull up resistors added as this was the norm, and the problem I see had gone un-noticed!

hope that hopes someone at somepoint!!

cheers
Andrew
 
Back
Top Bottom