MV-4F Address Errors Post Super Cap Mod

[firebrand]

Last week I pulled apart the NEO to swap my dead battery with a 1F 5.5v super cap and on re-assembly I started seeing an addressing error about 5 seconds into the attract mode of any/all games. Didn't seem to matter what slot I was loading.

I have version 3.1 of the Universe Bios so this data is coming from the bios error report screens.

Address Error
address: 0002A278
ac adrs: FFFFFFFF
r/w: read

d/a registers
D0: FFFFFFFF
D1: 00000000
D2: 0000FF66
D3: 00000040
D4: FFFFFFFF
D5: 00000000
D6: 00000108
D7: 000000AA

ssp: FFFF7EFD
sp: 2008

After enabling the hardware test on boot for the bios and was seeing this:

An error in your neo-geo system has been detected in the following region:
adrs: 00101802
wr: AAAA
re: AA8A

I reflashed the bios once to see if the problem resided with the bios since I've had some issues with the particular eprom that my first uni bios came on. It threw a few new errors and I was having some write failures on that eprom so I bought two HN27C1024 from ebay and re-wrote the 3.1 uni bios on one and I'm still seeing the two errors above.

Some additional info:
The button style sup caps I got did not have any polarity markings of any sort so I assumed they were symmetric and used my best judgement on installation after watching the mod videos. The system turns on and it is remembering my bios settings so it appears like the super cap is working, and the super cap hasn't exploded or anything yet sooo :-)

Any help would be greatly appreciated as I'm no expert on how to read the bios errors. Not sure how to merry them up to the circuits.

 

[Razoola]

This is an error in a RAM chip. You will need to replace the RAM to fix the error. In not sure where the RAM will be located on the PCB but others will know more than me.

Raz

 

[firebrand]

This is an error in a RAM chip. You will need to replace the RAM to fix the error. In not sure where the RAM will be located on the PCB but others will know more than me.

Raz

Thanks Raz, until I can figure out more I'll look around for cracked solder joints. Do you know if all of the RAM chips are on the daughter (top) board or can they be anywhere on either board?

 

[Xian Xi]

Most likely bad work ram, with the boards age I'd just swap both with new ones. I most times just replace all 4, no sense waiting for the others to fail. But good thing is that you can replace WRAM, BRAM, VRAM(slow) with the same type of ram. Of course only do it if you are good at smd repair.

 

[firebrand]

Most likely bad work ram, with the boards age I'd just swap both with new ones. I most times just replace all 4, no sense waiting for the others to fail. But good thing is that you can replace WRAM, BRAM, VRAM(slow) with the same type of ram. Of course only do it if you are good at smd repair.

Thanks Xian, I'll look around for some RAM chips. I'm pretty confident I can handle the desoldering/soldering job. Can those be found new, would places like DigiKey carry those or will I be surfing ebay or replacements?

Have you guys seen or contributed to this developer wiki?
https://wiki.neogeodev.org/index.php?title=Main_Page

 

[firebrand]

Xian, do you know if either of these chips will work for me?

I grabbed (what were listed as) HM62256BLFP-12T chips that turned out to be HM65256BLFP-12T chips. Before I get a refund I wanted to know if the HM65256BLFP-12T chips would work, the stats look similar on the datasheets but they are 65256 and not 62256.

If both are wrong that would be helpful to know as well, there are quite a few manufacturers making 62256 120NS chips in various packages. I couldn't find package info for the Sony chips that are currently on my board so I don't know if either of these packages will work.

HM62256BLFP-12T
450-mil, 28-pin, plastic SOP

HM65256BLFP-12T
??300-mil??, 28-pin, plastic SOP

http://pdf1.alldatasheet.com/datasheet-pdf/view/136264/HITACHI/HM65256BLFP-12T.html
http://www.datasheetarchive.com/dlmain/Datasheets-13/DSA-245852.pdf

 

[Xian Xi]

They're both 32kx8 SRAM. One is just a bigger body than the other, I'd honestly get something between 70-100. You should be fine with 120 though.

 

[firebrand]

They're both 32kx8 SRAM. One is just a bigger body than the other, I'd honestly get something between 70-100. You should be fine with 120 though.

Ok sounds good thanks! The seller did say he had a few other speeds in stock for the 62256 including 100NS and 150NS. Haven't received the order yet but I will chime back in with an update. If you think they look alright I might give em go unless the package size won't work with current circuit footprint.

I had gotten the 120NS target from the thread/post below, then I looked up the original SRAM chips on my NEO which are Sony CXK58256MM-12L (120NS).
channelmaniac had responded to the thread with the quote below.
http://www.neo-geo.com/forums/showt...VS-RAM-locations-a-preliminary-document/page2

For the RAMs:

The SMT (Surface Mount Technology) SRAM chips for Work, Backup, and Video SRAM are 32k x 8bit. Any JEDEC pin compatible 32k x 8bit SRAM IC will work as long as it is 120ns or faster speed.

This includes 62256, 43256, cxk58257, and others.

They MUST be SOIC type - not SOJ. SOJ has the leads turned under in a J shape and will NOT fit.

The 2k x 8 bit SRAM ICs used for the video SRAM are 2k x 8 bit JEDEC compliant pinout chips. Any chip of that type will work as long as they are 35ns or faster speed. This includes the CXK5814, CYC128A, and 6116 ICs, provided they are fast enough.

The 2k x 8 bit SRAM IC used for the sound SRAM is the same JEDEC type as the Video SRAM but is much slower. It must be 150ns or faster. The ultra fast 35ns ones that are used for the Video SRAM will work fine for a replacement but will be more expensive than a slow 150ns chip.

The 8k x 8 bit SRAM IC used for the Color SRAM (a.k.a. Palette RAM) is a JEDEC compliant SRAM IC. Any compliant SRAM IC of the same size and 120ns or faster will work. This includes ultra fast cache 64k RAM chips off of old scrap 386/486 boards.

 

[firebrand]

Ok, the seller is sending me the faster 100NS chips now, I'll let you know how it goes, thanks for all of the help thus far!

 

[firebrand]

It has been awhile. In my haste I had pulled a pad off while removing the first chip so I had to get a kit to bond a replacement pad. After rebonding, and doing a continuity test (positive) I pulled the second working ram chip and soldered on the two replacements... well... it got worse. Any ideas what I did now? Think it is salvageable? Here are some before and after shots.

I don't intend on giving up on this...

 

[firebrand]

I guess I'm going to have to add a logic probe to my growing EE toolset. These boards can be quite temperamental it seems. Close inspection of the traces and everything around the new SRAM looks clean, no bridging or cut traces of any kind, so I guess I did something wrong but I can't quite figure out what.

 

[pulstar]

Have you verified continuity with the new RAM you've put in or just that one pad that was lifted? If you haven't, it might be worth checking just to make sure it has been soldered in correctly.

 

[firebrand]

Have you verified continuity with the new RAM you've put in or just that one pad that was lifted? If you haven't, it might be worth checking just to make sure it has been soldered in correctly.

Would you recommend running a continuity check from the RAM all the way back to one of the chips? I'm a programmer by day and just an EE hobbyist so I know just enough to be dangerous, haha. Can I damage anything on the board by running a continuity check through the ICs or the RAM or would it be safer to get a logic probe? I know the check pushes a current through the probes to function so I usually just run the check on traces or non chip pins.

I just ran a continuity check on the pad before I epoxied over the trace patch on the replacement pad. The pad is from a CircuitMedic circuit frame, heat bonded. I need to get out the schematics and see where all of those traces go as the trace/circuit holes don't always work continuity, depends how thick the trace coating is.

The sadder thing is I bought a Japanese back-up MV4 on ebay to throw in my cab while I fiddle with my boards and it arrived not outputting Red. I know there isn't anything wrong with my CRT and the original owner had soldered his super gun right into the pinouts so now that is also sitting on the project table. :-P

 

[firebrand]

I got it booting, and it appears to be working!! No more address errors so far.

After some extensive amounts of testing, voltage, continuity and logic I tripled checked each of the ram pins and happened to noticed that pin 22 (OE) was floating, the pad and the solder had signals but the pin itself did not. A bit more flux and touching up the solder did the trick.

/wipes away the sweat

A BIG thanks to all of you guys for helping out!

I must say that I've been pouring over wiki.neogeodev.org looking for anything I might have missed and low and behold it was a solder joint, an important solder joint.

 

[firebrand]

Now back to the replacement MVS I bought to see if I can get the Red working, so I will have a working backup should something happen to my main board.