Hyper sports

[Lurch666]

Got an original hyper sports that was booting to a blank screen. The customs on the video PCB were corroded so I took them out and put them into turned pin sockets to preserve the legs.
Will also replace the corroded sockets with turned pin sockets in the PCB instead of normal sockets so I can plug these sockets into the ones in the PCB without damaging the socket.



Thanks to @Vamino I already have a Konami road blasters which has a near exact PCB so for testing I could use the customs from that PCB knowing they all work.
Once I have the hyper sports working I will swap back the original customs to see which ones are still fine.

Once I had known good customs in there I got a screen full of 0s but no other sign of life.
Next was testing the clock and reset and neither of these were present.
The Q and E clock signals that are required by the konami-1 CPU are generated on the video PCB from the custom at D1 but the reset is generated on the CPU PCB but passes to the video PCB then back to the CPU PCB. These signals go through the 244 at A1 and this is where they we failing so it looked like A1 was bad.

Now for some reason (either they had a surplus of 241s or a shortage of 244s) they used 241s to replace quite a few 244s on this PCB.
The main difference is the second enable on pin 19 is active high on a 241 where it's active low on a 244.
To cover this they cut pin 19 of the 241 so it wouldn't be held to GND and soldered it to pin 20 to hold it high.



Now if I wanted to replace this with a socket I would have a problem since I wanted to use 421s to keep things he same.
I ended up removing the bad chip,cutting the trace that connected pin 19 to pin 1 (pin 1 was connected to GND to hold it low) then after I had fitted the socket I connected pin 19 to pin 20 on the underside of the PCB. I could then drop an unmodified 241 into the socket and have it work normally.

Doing this I got reset and clock. but still wasn't getting any activity.
CPU roms checked out good but then I discovered the W/R were constantly high on the CPU ram.
Now the read/write signal is generated by the konami-1 CPU and also goes to the video PCB via another 244 at G1 on the video PCB and looking with the scope showed the signal was making to to G1 but not leaving it.
And since after the video PCB the read/write goes back to the CPU as write enable (Wren) that was the reason the ram wasn't active.
So that was another 241 instead of a 244 and replacing that got some activity.
The game was now booting but the screen was still all 0s. Sound was working though.
You can see the name entry screen here and the colours are moving but no proper graphics are showing:


So searching around I found little to no data activity on the rams at G9 and G10 on the video PCB.
This data comes via the 245 at H1 then goes to four chips (E9,E10,H10 and J10) and one was pulling the data signals down.
Since I had most issues with the chips buffering the connections between the PCBs I swapped out H1 first and this improved the signal but it was now squashed for some reason so I figured one of the other chips were affecting it.
Not wanting to shotgun all four chips I cut the D0 pin from H10 and J10 and checked the signal again.
No change so I soldered the legs back together.


And then tried cutting the D0 leg on E9 and that sorted the signal out.
Replacing that gave me this:



Getting there.
So now I hoped I could use the hypersports test rom to see if I could get any useful info from that.
It showed enough to tell me the G9 and G10 ram was bad.
I had already socketed these thinking they might be the cause of the data issue so I knew they were both good.
Then I found a poor address signal on pin 2.
This was down to a floating pin on the 157 at G6.
I piggybacked that and that cleared the doubled/garbled text.

Colours still bad and the background is still blue but I'm near to fixing this.
Need to replace the sockets for the customs and the graphics roms,replace the bad graphics roms (some have broken or badly corroded legs) and sort out the few remaining issues.

This has been an entertaining fix so far.
Thanks to @TheDaddy for sorting me out with a good deal for this PCB.

 

[Retroman839]

Good stuff !
Is it a faulty chip holding the signal low ?

 

[Lurch666]

Was in this case.

 

[Lurch666]

Forgot to mention the test rom was also showing the 2114 at H8 as bad so next job was replacing that.
Then the test rom showed the 2114 at H9 bad so replaced that as well.
Tried the test rom again and it was back to saying H8 was bad????

After some checking it turned out that the 2114 rams were fine and it was the LS245 at H10 that is the data bus transceiver for these chips was bad.

Replacing that got all the ram tests passed.

The incorrect colours were down to a bad graphics eprom at C14. Burned a new 27c128 and colours are good.
Text was missing a few pixels and that was a bad 27c128 at A14 so burned my last spare 128 and that's the text all good.

Got corrupted sprites and other graphics but that's all down to eproms with rusty legs.
Got a set of 27c128 on order from china and in the meantime will replace all the sockets since the ones already in this PCB are corroded to some extent.

Just need to figure out which mame files relate to which roms as in mame it's all files for 27c64 (8K) roms and that numbering is just c1 to c26.
On my PCB it uses 27c128 so I need to concatenate the two relevant 8K rom files to make a 16K rom file for the 27c128 roms this PCB uses.

I've managed to figure out some (the ones I burned to fix the text and colours) by reading the old eproms and figuring out from the contents which mame files I needed but the four roms at H17,H19,J17 and J19 are unreadable on my reader since it's not connecting to the legs properly so if anyone knows which files are which or has the 27c128 sized files for all the roms on this PCB that would help,thanks.

Just noticed in mame there are eight 8K files for gfx1 Which are text and background which match the four 16K roms in my PCB but only four 8K files for the sprites. My PCB has four 16K roms for sprites so either mame is missing data or mine isn't using the full rom.

 

[Lurch666]

Found a picture of the PCB and it looks like H17,H19,J17 and J19 are roms c11-c18 which are GFX1 in mame so I had it the wrong way round-GFX1 are the sprites.
Found some rewritable 256 roms that I will use in place of 128 eproms for experimentation to see if I can figure out the correct order.

So that leaves me with four GFX2 files of 8K for the four 128K roms for backgrounds and text but I managed to read most of those roms from my PCB so hopefully don't need the mame files.

 

[TheDaddy]

Out of interest did you use the slice to find the issue here :-

"After some checking it turned out that the 2114 rams were fine and it was the LS245 at H10 that is the data bus transceiver for these chips was bad.

Replacing that got all the ram tests passed."

Dave.

 

[Lurch666]

It helped.
Started by checking the 157s that handled the address lines with the slice and they all checked out,then I sliced the 245s that handled the data and the one at H10 was showing occasional small errors but most of the time it was passing the checks.
The 245 at J10 which handles the data for the other two 2114s checked out fine all the time so I suspected a problem with the one at H10 but the small amount of occasional errosr can sometimes be caused by a poor connection,poor GND or slightly different operating specs on the chips propagation.

But since it was the only chip that showed any errors I decided to swap it out and see what happened and the new chip fixed the issues.

 

[Lurch666]

So after looking at the files I managed to get from the roms I could read I figured out the order for all the sprite eproms:
H17,c12+c11
H19,c14+c13
G17,c15+c14
G19,c17+c16

Each 27c128 takes two 8K files from the mame hyperspt file but the order is reversed.

The bit that's confusing me is the other text/background roms.
The rom at C14 takes the mame files c22+c20 and the rom at A14 takes the mame files c26+c24 as they are both 27c128.
But the two roms at A12 and C12 are 27c64 and there no more GFX files in the mame hypersports zip to go into them.
I ended up using two 27c64 erpoms from a bootleg hypersports (that I still haven't burned the prom for) and they appear to work.

So that's all the sprites and graphics sorted but then a (?)last intermittent issue appeared.
All the sprites started appearing only on the left hand side of the screen. They would move up and down but were stuck on the far left.
This would go away when I was trying to figure out what the issue was then come back later so I suspect a chip near failing.

I'm guessing it's got something to do with the XPOSI signal coming from pin 8 of F12 that goes into the 273 at E12 pin 11.
But at the moment the sprites are behaving themselves so until it messes up again there's nothing for me to check.

 

[Lurch666]

That took a while.
Turns out it was the 273 at C16.
Didn't figure it out in any technical way since what I thought would cause the issue wasn't actually faulty.

Ended up just shotgunning the sprite section with my slice,pretty sure I had checked that chip in earlier tests but now it was showing as bad.
When I started to desolder it I could see why the issue was intermittent:


Two legs were corroded to the point of failing. I assume sometimes they were just about making a connection and the rest of the time (when the sprites were stuck to the left) they were disconnected.
Replacing that appears to have fixed the issue but since it was intermittent will have to keep checking to see if the issue rear it's head again.