MVS SLOT4FS graphics problem with games and mem viewer

maki

Edo Express Delivery Guy
Joined
Jan 1, 2022
Posts
332
Hi,

have a MVS SLOT4FS here that passes the basic DIAG Bios checks, it has no sound and massive glitches when displaying either the mem viewer or running games.
Everything looks perfectly fine without game and in the cross hatch/menu etc, still no sound though.

Happens on all slots, didn't notice a difference between the 4

Any hints where to look first?

PXL_20230330_124604772.jpg


PXL_20230330_131219851.jpg

PXL_20230330_131152036.jpg
 
Last edited:

ack

Ninja Combat Warrior
15 Year Member
Joined
Apr 9, 2009
Posts
538
Never seen an issue like that before, but would suspect something with one of the custom graphic chips on the cpu board.
 

maki

Edo Express Delivery Guy
Joined
Jan 1, 2022
Posts
332
Thank you @ack , I really appreciate your answer :)

TBH I like that answer, much more than testing/swapping one of the many G0/244/253 on the top board.
This is the FS version, with the 2nd gen chipset (LSPC2-A2, NEO-B1), plenty of those around here.

I was thinking maybe it is the ZMC2 on the mainboard as the C-ROM data is handled by it, while all slots seems very similar in terms of graphics error, there is slight differences, the picture of the memory viewer is from the first slot, it won't start (black screen ) on any other slot.

I'll swap the LSPC2-A2 first, will report back.

here is a video of Sam Showdown in slot 1:

here it is in slot 4:
 

maki

Edo Express Delivery Guy
Joined
Jan 1, 2022
Posts
332
I've swapped the NEO-B1 and LSPC2-A2 few times, turns out lots of them were dead, with working ones, no change.
Then I've swapped the NEO-ZMC2, still no change.

PXL_20230401_071709442.jpg

I was surprised to find how many of them were defunct in my "spare" pile (turns out its a junk pile :) ), with some of them I could reproduce the errors I got on the donor boards, dark screen only with faint red color on top or click or death.

So back to the top board then, I still think the defect is in a place that must affect all slots equally, like the output of the multiplexer that goes to the main board.
The games run, so not all data is affected?

On the top board there is five NEO-253 on the right, on the left there is six NEO-244 and four NEO-G0.

Just by the numbers, the five 253 and the six 244 look like candidates for "places that all data has to pass through".

Any guess which one is the output of the multiplexer?
That one should be directly connected to the ports right?

PXL_20230401_072118883.jpg

Edit:
only thing similar I've found is this video of a broken MV4FS

It's also from a MV4FS board, almost identical graphic problem but the fix layer works, unlike on mine.
Im thinking its a couple of defects on my board that add up to this result.

Edit 2:

so I swapped the FIX an LO ROM, no change
replaced the NEO-I0 with another one, got worse, even the text in the cross hatch was garbled, put back the original as I do not have another one

since I do not have any spare for the 253 and the 244, I'm can't replace them, but I sure can swap the ones that are there..

swapped the 244 between A1 and A2, no change

swapped the 253 between E5 and D5, now the slight difference in the graphics error between the 1st and the 4th slot are swapped as well, so I got that at least, will swap other 253 into these spots

those 253 are hard to find, only in second revision 4 slot PCBs.. bummer
on the bright side they're just 74HC253 packaged together
 
Last edited:

ack

Ninja Combat Warrior
15 Year Member
Joined
Apr 9, 2009
Posts
538
What does the diag bios look like?
 

maki

Edo Express Delivery Guy
Joined
Jan 1, 2022
Posts
332
the diag bios looks fine, same with the crosshatch and the test menu

PXL_20230401_181201994.jpg

been swapping more ASICs, no real change so far (minor differences, white dotted patterns vs solid black background), only thing that made an obvious difference was the NEO-I0, not everything was worse with the other defective one, some of the right letters came through, at boot I could recognise "SNK" in blue but it was stretched and the screen was garbled in a different way
 

maki

Edo Express Delivery Guy
Joined
Jan 1, 2022
Posts
332
Progress!

Swapped two 244:
C2 <-> D2

C2: Buffers for P bus lines
D2: Buffers for P bus and other lines

source:

now I can see the memory viewer, however games don't start anymore

slot 1
PXL_20230401_205008904.jpg

slot 2
PXL_20230401_205745068.jpg

slot 3
PXL_20230401_205706269.jpg

slot 4
PXL_20230401_205607039.jpg


seems like swapping "sacrificed" P bus for Graphics addresses or similar, need to ave a closer look at what they do in these positions
 
Last edited:
  • Like
Reactions: ack

maki

Edo Express Delivery Guy
Joined
Jan 1, 2022
Posts
332
I've probed the 244 in D2, pins 61 - 63 are dead, OUT16 and OUT17 are the buffers for A18 and A19
swapped that one with the 244 in E2 (= "Buffers for ADPCM lines", no audio currently anyways)

got something working here :)
PXL_20230402_074354101.jpg
PXL_20230402_074334041.jpg

still some minor issues in the graphics and no sound at all, without some spares its going to be really hard

I've ordered some 74LS244 and 74HCT244, since there is no NEO-244 available, I'll need to "jump" the defunct parts of the NEO-244 and replace it with discrete 74244
its the next best thing to a real NEO-244, could get ugly though

if you're in Europe and want to throw a few NEO-244 at me don't hold back ;)

found this very helpful page about the first gen of the MV4, it lists the discrete used ICs that go to the connectors:

working assumption is that the second gen just integrated all these discrete ICs, so I can rebuild whats missing/defunct

bummer that there ain't a drop-in replacement for the 244, like furrteks 273 and PCM spares

Edit:
the original location of the defunct 244 was 2C = Buffers for P bus lines
so P14 and P15 were dead in the water, potentially more

thats what caused the weird graphics and missing text, or at least part of it

 
Last edited:

maki

Edo Express Delivery Guy
Joined
Jan 1, 2022
Posts
332
To conclude this in case someone else stumbles on it:
- the NEO-244 at position C2 ( Buffer for P Bus) had half of it buffers shot - this caused the graphics problems
- the NEO-244 at position E2 (Buffers for ADPCM) had 4 input/8 output shot, this caused missing sound and probably instability, MS4 would hang in the demo mode at the same spot
- for the NEO-244 at position A2 (Buffers for control lines) you really want a perfectly working NEO-244, this one deals with the clocks, including the master clock with 24MHz, I wouldn't risk running it through patch wires with my routing skills

Lots of defektive outputs, sometimes just a single one of the two, I still replaced the buffer as such. Most of times both outputs would be defective, in one case the input turned into a an output..

FWIW, I wasn't able to check high frequency signals on A2 with my low quality logic probe, an old analog scope was much better for that.

I've used 74HCT244 SMD replacements on breakout boards, can't say for sure but I'm assuming the NEO-244 used LS, but HCT should be compatible.

two 244 for C2:
Each input is going to two outputs, each output is assigned to two slots. (1&2 and 3&4).
I've tried using a single 244, one input goes to one output and just wire that output to all slots, this would save half of the 244 needed to replace a NEO-244 (which has 5), but the risk is that the noise from 4 instead of 2 slots might cause instability.

I've stacked two breakout boards with an 244 each, connected the inputs (and GND, +5V, /OE) via pins, and routed each output separately to keep them separated.
(ignore the 3 long red wires, thats a spare buffer, I'll keep the wires until I'm certain I won't need it)

PXL_20230407_225825348.jpg

PXL_20230410_081427852.jpg

The bling-bling gold is a failed attempt to distract from the wiring mess.. but using same colors for each input helps to keep an overview.

another 244 for E2:
only needed to replace 4 input/8 outputs, a single 244 would do, also connected two inputs here (jumpers)

PXL_20230410_081333395.jpg

Graphics are fine now, sound works mostly now (still a minor issue on slots 3&4, I'll fix that later).

Replacing the NEO-244 with discrete logic ICs is a spacing/routing problem, replacing a single NEO- 244 means stacking 5 breakout boards with SMD ICs is possible but fugly, there would be nearly 60 wires coming out of that..

TBH, right now I'm not that confident about this machine, kind of surprised by the amount of failed 244s, I'll keep testing it until I'm more confident or it fails.

I'm not aware of any drop-in replacements for NEO-244, only way to get the original replacement parts to cannibalise other MV4F/MV4FS.
 
  • Like
Reactions: ack
Top